entire volume - Casualty Actuarial Society

Transcription

VOLUME
LXXX11
NUMBERS
156 AND
157
PROCEEDINGS
OF THE
Casualty Actuarial Society
19 14
ORGANIZED
1995
VOLUME
LXXX11
Number 156-May
Number 157-November
1995
1995
COPYRIGHT--l996
CASUALTY
ACTUARIAL
SOCIETY
ALL RIGHTS RESERVED
Lkmq
of Congress Caralog No. HG995h C3
ISSN 0891-1980
FOREWORD
Actuarial science originated in England in 1792 in the early days of life insurance.
Because of the technical nature of the business, the first actuaries were mathematicians.
Eventually, their numerical growth resulted in the formation of the Institute of Actuaries
in England in l&18. Eight years later, in Scotland, the Faculty of Actuaries was formed.
In the United States, the Actuarial Society of America was formed in 1%X9 and the
American Institute of Actuaries in 1909. These lwo American organtiations merged in
1949 to become the Society of Actuaries.
In the early years of the 20th Century in the United States, problems requiring actuarial treatment were emerging in sickness, disability, and casualty insurance-particularly
in workers compensation, which was introduced in 1911. The differences between the
new problems and those of traditional life insurance led to the organization of the the
Casualty Actuarial and Statistical Society of America in 1914. Dr. I. M. Rubinow, who
was responsible for the Society’s formation, bccamc its first president. At the time of
its Cormation, the Casualty Actuarial and Statistical Society of America had 97 charter
mcmbcrs of the grade of Fellow. The Society adopted its present name, the Casualty
Actuarial Society, on May 14, 1921.
The purpose of the Society is to advance the body of knowledge of actuarial science in
applications other than life insurance, to establish and maintain standards of qualification
for membership, to promote and maintain high standards of conduct and competence for
the mcmbcrs, and to increase the awareness of actuarial science. The Society’s activities
in support of this purpose include communication with those affcctcd by insurance, presentation and discussion of papers, attendance at seminars and workshops, collection of
a library, research, and other means.
Since the problems of workers compensation were the most urgent at the time of the
Society’s formation, many of the Society’s original members played a leading part in dcvcloping the scientific basis for that line of insurance. From the beginning, however, the
Society has grown constantly, not only in membership, but also in range of interest and in
scientific and related contributions to all lines of insurance other than life, including automobile, liability other than automobile, Circ, homeowners, commercial multiple peril, and
others. Thcsc contributions are found principally in original papers prepared by members
of the Society and published annually in the Proceedings of fhe Casualty Acfuarial Sociefy. The presidential addresses, also published in the Proceedings, have called attention
to the most pressing actuarial problems, some of them still unsolved, that have [aced the
industry over the years.
The membership of the Society includes actuaries employed by insurance companies,
industry advisory orga&ations,
national brokers, accounting firms, educational institutions, state insurance departments, and the federal govcrnmcnt. It also includes independent consultants. The Society has two classes of members, Fellows and Associates.
Both classes require successful completion of examinations, held in the spring and fall of
each year in various cities of the United States, Canada, Bermuda, and selected overseas
sites. In addition, Associateship requires completion of the CAS Course on Professionalism.
The publications of the Society and their respective prices arc listed in the Society’s
Yearbook. The Syllabus of Examinations outlines the course of study recommended for
the examinations. Both the Yearbook, at a charge of $40 (U.S.), and the Syllabus of Examinafions, without charge, may be obtained from the Casualty Actuarial Society, 1100
North Glcbc Road, Suite 600, Arlington, Virginia 22201.
I
JANUARY I, 1995
EXECUTIVE COUNCIL*
ALLAN M. KAUFMAN . . . , , . .
..
. . . . President
ALBERT J. BEER
....
...
. . . . . President-Elect
PAULBRAITHWAITE
. . . . . . .
. Vice President-Administration
JOHN J. KOLLAR
. . . . . . . _ . . . . . . . . . Vice President-Admissions
DAVID N. HAWING
. . . . . . . . . Vice President-Continuing Education
ALICE H. CANNON. . Vice President-Pmgram.r and Communications
MICHAELJ. MILLER . . . . Vice President-Research and Development
THE BOARD OF DIRECTORS
ALLAN M.
ALBERT J.
Imtni~difltr
BEER
PHILLIP
President
. . . . . . . . . . President-Elect
Post Presirlentt
IRENEK.BASS
Ekttd
........
.......
KAUFMAN
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Dirwtorst
N. BEN-ZVI
CLAUDETTE
CANTIN
.
.
BRIAN E. Scorn . . .
SUSANT. SZKODA
STWEN F. GOLDBERG
PATRICK J. GRANNAN
ANNE E. KELLEY
ROBERT
..I994
.
...
.
..
.
.
.
S. MICCOLIS
ROBERT F. CONGER
JOHNM. PURPLE
RICHARDH. SNADER
KEVIN B. THOMPSON
1995
1995
1995
I995
1996
1996
1996
I996
I997
1997
1997
1997
* Term expires at the 199.5Annual Meeting. All membersof the Executive
Council itre Officers. The Vice President-Administration also serves as
Secretaryand Treasurer.
7 Term expires at Annual Meeting of year given.
II
1995 PROCEEDINGS
CONTENTS OF VOLUME LXXX11
Page
ADDRESS
15, 1995
TO NEW MEMBERS-MAY
David G. Hartman . .
MINUTES
OF THE MAY
PAPERS PRESENTED
1995
...............
MEETING
. . . . . . . . . .
AT THE NOVEMBER
1995 MEETING
.
........ I
. .
7
Balancing Transaction Costs and Risk Load in Risk Sharing
Arrangements
Clive L. Keatinge . . . . . . . . . . .
. . . . . . . . . . . . . . .23
DISCUSSION
OF PAPER PUBLISHED
IN VOLUME
LXXIV
A Note on the Gap Between Target and Expected Underwriting
Profit Margins
Emilio C. Venezian (November 1987)
Discussion by William R. Gillam .
. . . . . . . . . . . . . 56
DISCUSSION
OF PAPER PUBLISHED
IN VOLUME
LXXX1
Unbiased Loss Development Factors
Daniel M. Murphy (November 1994)
Discussion by Daniel F. Gogol, Ph.D. . . . . . . . . . . . . . . . . 72
DKLSSION
OF PAPER PUBLISHED
IN VOLUME
LXXVII
Risk Loads for Insurers
Sholom Feldblum (November 1990)
Discussion by Stephen W. Philbrick (May 1991)
Author’s Reply to Discussion (November 1993)
Discussion by Todd R. Bault . . . . . . . . . . . . . . . . . . . . . . .78
Author’s Reply to Discussion
. . . . . . . . . . . . . . , . . . . 97
DISCUSSION
OF PAPER PUBLISHED
IN VOLUME
LXX11
A Simulation Test of Prediction Errors of Loss Reserve
Estimation Techniques
James N. Stanard (May 1985)
Discussion by Edward F. Peck
. . . . . . . . . . . . . . . . . 104
111
1995 PROCEEDINGS
CONTENTS OF VOLUME LXXX11
Page
ADDRESS
13, 1995
TO NEW MEMBF.RS-NOVI~M~~I:R
121
Michael A. Walters . . . . . .
PRIXDENTIAI.
ADDRESS-N•
VEMIWR
13.
1995
126
Allan M. Kaufman
MINUTES
OF THE NOVEMBER
REPORT OF THE VICE
FIN.AN(.I.~L.
REf’OKr
1995 METING
PRESIDENT-ADMINISTRATION
.
...........
C!,NI)IIWf:s
160
167
...................................
1995 EXA~JINA-IIONS-SUCCESSFUL
138
.............
168
OBITWRIIZS
William J. Hazam ................................
Laura J. Manley ..................................
Kenneth L. McIntosh ..............................
Robert W. Parlin .................................
James W. Thomas ...............................
INDEX
TO VOLUME
LXXX11 ............................
.206
209
211
,212
,213
2 14
NOTICE
Papers submitted to the Procwdbzgs
of’ t/w Cmdty
Actuarial
Sockty are
subject to review by the members of the Committee
on Review of Papers and,
where appropriate.
additional individuals
with expertise in the relevant topics. In
order 10 qualify for publication,
a paper must be relevant to casualty actuarial
science. include original research ideas and/or techniques, or have special educational value, and must not have been previously
copyrighted
or published
or he concurrently
considered
for publication
elsewhere. Specific instructions
for preparation
and submission of papers are included in the Ycarhod
of the
(‘a\ualty
Actuarial
Society.
The Society
clc\. criticisms,
is not responsible for statements of opinion cxprc\aed
and discussions publijhcd
in thc$c I’~oc,c,c,dir7,~.\.
IV
in the arti-
Volume LXXXII,
Part 1
No. 156
PROCEEDINGS
May 14,15,16,17,1995
ADDRESS TO NEW MEMBERS-MAY
15, 1995
DAVID G. HARTMAN
Isn’t it appropriate that we are here today in the shadow of
Gateway Arch as you new Fellows and new Associates are at the
gateway of the rest of your career! What I would like to do today
is talk to you in three areas: First, compliment you, with both an
cci” and an “e”; second, share with you some of the experiences
of my contemporaries; and third, chullen~e you.
ComplimentlComplement
First of all, the compliment. Congratulations to you on a job
well done! To Fellows, for making it all the way through the exams; to Associates, for making it at least 70% of the way through.
We appreciate very much the stick-to-itiveness that you have
demonstrated and the sacrifice that this has meant. Keep at it.
Compliments also to spouses who have been so supportive of you
as you have taken the exams, as well as to friends, co-workers,
and others who have been supportive of you in attaining this particular milestone. You have now entered a profession.
1
2
ADDRESS
TO NEW MEMBERS
There are four hallmarks to a profession. First, a profession
has a core of basic knowledge unique to that profession. Second, a professional continues his or her education. Both of these
functions are performed in the actuarial profession by the Casualty Actuarial Society or, for our life, pension and health counterparts, the Society of Actuaries. To complement that, with an
“e,” is the American Academy of Actuaries, primarily performing two other key parts of professionalism. The third hallmark
of a profession is that there are standards. One standard would
be a Code of Conduct; others would be the Standards of Practice as promulgated by the Actuarial Standards Board (ASB).
The fourth hallmark is discipline, which is performed by the Actuarial Board for Counseling and Discipline (ABCD). They like
to emphasize that counseling is the key part of their function. I
urge you to complement your CAS membership by becoming a
member of the Academy, if you have not already done so.
Con tempera ries
Moving on to the area of contemporaries, I thought it would
be interesting for you to let me share with you the paths taken
by the six of us who became Fellows together at the May 1972
meeting. Unfortunately, one of those, Edward R. Murray, died
in 1984. I’d like to give you a thumbnail sketch of the other
five of us and hopefully have some of our experiences stimulate thoughts about your own career. To set the stage, I’d like to
recall for you a definition of an actuary that past CAS President
Fred Kilbourne presented back in 1978. That definition is that
“an actuary is that professional who is trained in evaluating the
current financial implications of future contingent events.”
Walt Stewart and I are, more or less, traditional company
actuaries. Walt is vice president and senior actuary of CIGNA
Property and Casualty. He has responsibility for research, pricing
policy, claims operation support, workers compensation, administration, the actuarial student program, statistical reporting, and
part of actuarial systems. Walt has been a representative of his
ADDRESS
TO NEW MEMUERS
3
company on a number of industry committees, principally at IS0
and the National Council on Compensation Insurance. Some of
the research that he has brought forward has been very much
the kind of thing that would make him an actuary’s actuary.
As indicated in my introduction, I am managing director and
senior vice president and chief actuary at the Chubb Group of
Insurance Companies. In addition to my many professional activities, I am or have been active in leadership of church, Scouts,
United Way, and local hospital groups.
Bob Anker is, and has been for the past three years, president
and chief operating officer of Lincoln National Corporation, an
insurance holding company including American States Insurance
Companies and the Lincoln National Life Insurance Company
of which he is chairman and CEO-a casualty actuary serving as
president of a life company! Bob worked as a casualty actuary
for ten years at Employers Insurance of Wausau prior to joining
American States in 1974. As an example of his continuing education, he has earned both the CPCU and the CLU designations.
He has served the profession as vice president of development
and director of the Casualty Actuarial Society and is currently a
director of the American Academy of Actuaries. He has written
reviews of papers and frequently served as a panelist on CAS
meeting programs. He is a past president of the Midwestern Actuarial Forum. Looking at community service, Bob has served, or
is serving, on the boards of 11 civic organizations.
Bob Bergen is chairman of Investment Research Institute, Incorporated, an organization he co-founded in 1981 with another
CAS member, Bernie Shaeffer. IRI publishes the Option Advisor and Fund Profit Alert newsletters with over 20,000 subscribers. These newsletters provide investment advice in the
stock, option, and mutual fund areas. Bob worked for IS0 for ten
years and for Great American for seven years prior to forming
IRI. Bob has written me saying that his community service encompasses involvement with youth sports, including sponsoring
youth baseball teams. consulting on league insurance programs,
4
ADDRESS
TO NEW MEMBERS
fundraising, and handling team statistician chores. Bob says that
while the financial markets are a natural extension of his interest
in economics, statistics, and just plain numbers, the disciplines of
the CAS have served him well in many aspects of a growing business. Attention to detail and ability to interpret financial data are
valuable in the creation, planning, and analysis of marketing programs. The ability to analyze statistical information for accuracy
and relevance is important in many areas, especially for management reports and accounting and financial statements, as well as
in research relating to deriving stock and option recommendations for subscribers.
Charlie Rinehart is chairman of the board and chief executive officer of Home Savings of America, which is the largest
savings and loan in the United States with $54 billion of assets.
Charlie worked for Fireman’s Fund for 15 years where he rose to
the number two job in the Actuarial Department in his first six
years, then managed the Personal Lines Underwriting Department, and then the Commercial Group Department, which did
mass merchandising. He was in charge of all field operations before he left in 1983 to become president of Avco Finance. He
joined Home Savings in 1989. Charlie tells me that he finds that
there are many similarities between his work in property/casualty
insurance and consumer finance. The pricing of consumer finance involves predicting default risk by first reviewing historical data of homogeneous groupings and also by reviewing trends
in external economic factors, such as home prices and the job
market. Both involve setting reserves using operations research
and econometric modeling tools. Both involve financial statement
analysis. He found the only real adjustment between the two to
be what comes first. In insurance, what comes first is the payment of a premium which is set up as an asset, and then liabilities are established for unearned premium reserves and loss
reserves. In consumer finance, on the other hand, first money is
borrowed establishing a liability, and then assets are established
in the way of mortgages.
So all five of us are actuaries in terms of the definition that
Fred Kilbourne enunciated, yet our training has taken us into
pricing, reserving and financial analysis in property and casualty,
life, consumer finance, and investment applications. As you look
at your future, I hope you will consider Fred’s definition as a
broad one encompassing traditional and non-traditional actuarial
roles for yourself.
Challenges
Moving on now to the area of challenges as you step through
this gateway, let me offer several specific items for your consideration.
First, do your best at what you are doing right now. As the
old adage says, if you rest on your laurels, you’re wearing them
in the wrong place. You have received training as a casualty actuary, and that is something special throughout the world. Nowhere
else in the world does any group of actuaries receive such specialized training as you are receiving in the practice of casualty
insurance, or what is referred to as general insurance outside the
U.S. As you do your best, make sure to add value to the process
in which you are involved. Your first priority on the job is to provide your customer with what they need. Charlie Rinehart captured that by saying, “find solutions’for the business problems.”
An elegant mathematical model without practical application is
unlikely to add value. If you do your current job well, you will be
building a solid foundation for the future.
Second, continue your education. For those of you who are
new Associates at this meeting, I urge you to finish your Fellowship exams. Never stop growing. I look at myself years ago
when I was deciding whether I should be a life actuary or a casualty actuary, and I deliberately chose the casualty field because
it was an area where things were always changing. You could
never learn all there is to know about the practice of casualty
actuarial science. Keep ahead of the curve as you continue your
6
ADDRESS
'TO NEW MEMBERS
education. The changes in the areas of legal, economic, and technological developments seem to be occurring at an exponential
pace. Think of how your training and your skills can be applied
to keep pace with, or even stay ahead of, these developments.
Third, maintain the highest ethical standards in what you do.
In short, abide by our Code of Conduct. A key characteristic of
a professional, or any leader, is integrity. You can work for years
building a reputation and then have it dashed forever with one
ethical violation. Mistakes can be forgiven. but a lapse in ethical
behavior cannot be.
These three challenges I’ve given you correspond very closely
with the first three hallmarks of a profession. First, having and
applying a base of knowledge unique to the profession; second,
continuing your education; and third, abiding by standards. The
fourth challenge I am going to give is not directly related to discipline, which is the fourth hallmark of the profession, but it does
require discipline in the way you conduct your life. The advice
or the challenge I give to you is to volunteer. Make sure to make
time to help others; to give back a part of what you have received. Be of service to both your profession and to your community. You are aware that back in 1989, and also in April of this
year, the Jobs Rated Almanac rated the job of actuary as number
one. As such, we are really blessed. Hopefully. you will be able
to share this blessing by volunteering to serve on a CAS or an
Academy committee, or to put a face on our profession in your
community.
In summary, it is my pleasure to welcome you to the actuarial
profession. I challenge you to be seen as a professional. You’ve
earned it: now maintain and enhance it.
MINUTES OF THE 1995 SPRING MEETING
May 14-17, 1995
THE ADAM’S
MARK
HOTEL,
ST. LOUIS,
MISSOURI
Sundaq; May 14, I995
The Board of Directors held their regular quarterly meeting
from noon to 5:00 p.m.
Registration was held from 4:00 p.m. to 6:00 p.m.
From 5:30 p.m. to 6:30 p.m., there was a special presentation to
new Associates and their guests. The session included an introduction to the standards of professional conduct and the CAS committee structure.
A welcome reception for all members and guests was held from
6:30 p.m. to 7:30 p.m.
Monday,
May 15, I995
Registration continued from 7:00 a.m. to 8:00 a.m.
CAS President Allan M. Kaufman opened the meeting at 8:00
a.m. and recognized special guest Hans Btihlmann, Professor of
Mathematics, Swiss Federal Institute of Technology and the first
Bowles Chair of Actuarial Science at Georgia State University.
Professor Biihlmann participated in the general session on Dynamic Financial Analysis.
Mr. Kaufman also recognized past presidents of the CAS who
were in attendance at the meeting, including Irene K. Bass (1994)
Phillip N. Ben-Zvi (1985), Ronald L. Bornhuetter (1975), Charles
A. Bryan (1990), Michael Fusco (1989), David G. Hartman
(1987), W. James MacGinnitie (1979), JeromcA. Scheibl (1980)
Michael L. Toothman (1991), and Michael A. Walters (1986).
7
8
MINUTES OFTHE
IV5 SPKINC; MEETING
Paul Braithwaite, David Hafling, John Kollar, and Michael
Miller announced the 123 new Associates and the 17 new Fellows.
The names of these individuals follow.
NEW FELLOWS
Timothy JamesCremin Mark Priven
EduardJ. Pulkstenis
Bradley A. Granger
Craig W. Kliethermes John F. Rathgeber
Mathieu Lamy
David M. Savage
SuzanneMartin
Jeffrey Jay Scott
Russell Steingiser
Brett Evan Miller
Eileen M. Sweeney
Yuan-YuanTang
ThomasC. Tote
John Vincent
Van de Water
PeterGerardWick
NEW ASSOCIATES
Rimma Abian
Christopher R. Allan
John Porter Alltop
K. Athula P.Alwis
StevenDouglas
Armstrong
Martin Scott Arnold
Bruce J. Bergeron
StevenLouis Berman
Corey J. Bilot
Carol Ann Blomstrom
John T. Bonsignore
Douglas J. Bradac
Kevin Michael Brady
Betsy A. Branagan
JamesL. Bresnahan
Lisa Jenny Brubaker
Elliot R. Burn
Tara Elizabeth Bush
J’ne Elizabeth
Byckovski
SandraL. Cagley
PamelaJ. Cagney
DouglasAnthony
Carlone
Martin Carrier
Jill Christine Cecchini
HeatherLee Chalfant
Jean-Francois
Chalifoux
PeggyCheng
Gary C. K. Cheung
ChristopherJohn
Claus
William FrancisCosta
Christopher George
Cunniff
SeanRichard Devlin
BehramMehelli
Dinshaw
William A. Dowel1
Kimberly J. Drennan
Pierre Drolet
StephenC. Dugan
Tammy Lynn Dye
Jeffrey Eddinger
SvenAnders Ericson
JamesG. Evans
StevenJ. Finkelstein
Daniel JosephFlick
Andre F. Fontaine
SusanTereseGarnier
Christopher H. Geering
Eric J. Gesick
John Thomas Gleba
John Edmund Green
StevenA. Green
Charles R. Grilliot
Julie Kay Halper
David S. Harris
Betty-Jo Hill
John V. Hinton
JasonN. Hoffman
John Frederick
Huddleston
Li Hwan Hwang
MlNUTESOFTHE1995SPRINGMEETING
Brian L. Ingle
Christian Jobidon
Daniel Keith Johnson
Gail E. Kappeler
Lowell J. Keith
Thomas Paul Kenia
Michael Benjamin
Kessler
Jean-Raymond
Kingsley
Gary R. Kratzen
Brian Scott Krick
Marc La Palme
Debra K. Larcher
Gregory D. Larcher
Daniel EugeneLents
Edward A. Lindsay
Richard Borge Lord
Cornwell H. Mah
Anthony Leroy
Manzitto
Scott Andrew Martin
Tracey Lynn Matthew
Camley A. Mazloom
Deborah Lynn
McCrary
Mike McCutchan
Kelly S. McKeethan
Lynne Sener
McWithey
Claus Siegfried
Metzner
Paul W. Mills
Anne Hoban Moore
KennethBowers
Morgan, Jr.
Kevin T. Murphy
Hiep Trong Nguyen
JamesL. Nutting
Milary NadeanOlson
ThomasPassante
Nicholas H. Pastor
Claude Penland
William Peter
Genevi&vePineau
Robert Emmett
QuaneIII
PeterSebastianRauner
Natalie J. Rekittke
Scott Reynolds
Meredith G.
Richardson
9
John Walter Rollins
RajeshV.
Sahasrabuddhe
Christina Lee Scannell
Marilyn Schafer
Michael Jeffrey Scholl
Mary Kathryn Smith
John B. Sopkowicz
Michael J. Sperduto
Scott D. Spurgat
Scott Timothy Stelljes
Kevin D. Strous
StevenJ. Symon
Joy Yukiko Takahashi
David Michael Tern6
Daniel A. Tess
Son Trong Tu
Eric Vaith
Cynthia Leigh Vidal
Robert J. Walling III
StevenBoyce White
Elizabeth R. Wiesner
Michael J. Williams
Mr. Kaufman then introduced David G. Hartman, a past president of the Society, who presented the Address to New Members.
Alice H. Gannon, CAS Vice President of Programs and Communications, presented the highlights of the program.
It was announced that no Proceedings papers would be presented at this meeting. In response to a CAS call for papers on Dynamic Financial Analysis, it was announced that eight discussion
papers would be presented at the meeting, and bound in the 1995
10
Sflh’l’TES OFTHE 1’295 SPRING hlEETING
CAS Discussion Paper Program, titled “Incorporating
tors in Dynamic Financial Analysis”.
Risk Fac-
Mr. Kaufman then began the presentation of awards. He gave
some background information about the Charles A. Hachemeister
Prize and announced the prize winners, Michel Laparra, Isabelle
Lion, and Christian Partrat, all from France. The award will be officially presented at the 1995 CAS Annual Meeting in San Diego,
California, in November 1995.
Mr. Kaufman also announced that Larry Lickteig is the recipient of the 1995 Harold W. Schloss Memorial Scholarship Fund.
He will be presented with a $500 scholarship.
Patrick J. Grannan spoke to the CAS membership about the activities of the American Academy of Actuaries Committee on
Property/Liability
Financial Reporting Issues. of which he is the
chairperson.
Mr. Kaufman then concluded the business session of the Spring
Meeting.
After a refreshment break, Albert J. Beer introduced R. W. Apple, Washington Bureau Chief for the New York Times. Mr. Apple
was the keynote speaker and spoke to the CAS membership about
the first 100 days of the new Republican Congress.
The first general session was held from IO:45 a.m. to 12: 15
p.m.:
“Dynamic Financial Analysis”
Moderator: David G. Hartman
Senior Vice President and Chief Actuary
Chubb Group of Insurance Companies
Panelists: Hans Btihlmann
Professor of Mathematics
Swiss Federal Institute of Technology
Bowles Chair of Actuarial Science
Georgia State University
Shane A. Chalke
Chief Executive Officer
CHALKE Incorporated
Weston Hicks
Analyst
Sanford C. Bernstein & Co.
After a luncheon, the afternoon was devoted to concurrent sessions which included a presentation of the discussion papers and
panel presentations.
The discussion papers presented were:
1. “Mean-Variance Analysis and the Diversification of Risk”
Author:
Leigh J. Halliwell
Louisiana Workers’ Compensation
Corporation
I.3 “How to Best Use Engineering Risk Analysis Models and
Geographic Information Systems to Assess Financial Risk
from Hurricanes”
Authors:
Auguste Boissonnade, Ph.D.
Risk Management Solutions, Inc.
Peter Ulrich
Risk Management Solutions. Inc.
3.
4.
“Measuring and Managing Catastrophe Risk’
Authors:
Ronald T. Kozlowski
Tillinghast/Towers Perrin
Stuart B. Mathewson
TillinghastlTowers Perrin
“Managing Catastrophe Risk”
Author:
Glenn G. Meyers, Ph.D.
Insurance Services Office, Inc.
MINUTES
12
5.
6.
7.
8.
OFTHE
1995 SPRING MEETING
“An Approach to Evaluating Asset Allocation Strategies
for Property/Casualty Insurance Companies”
Manuel Almagro, Jr.
Authors:
TillinghastiTowers Perrin
Stephen M. Sonlin
“New Products-Uncertainty
of Cost, Measurement and
Control of Risks, and Implied Profit Margins”
Owen M. Gleeson
Author:
President
Financial Analysis and Control Systems
“A Decade of Cash Flow Testing-Some Lessons
Learned”
Ralph S. Blanchard III
Author:
Aetna Life & Casualty
Eduardo P. Marchena
Aetna Life & Casualty
“Forecasting the Future: Stochastic Simulation and
Scenario Testing”
Sholom Feldblum
Author:
Liberty Mutual Insurance Company
The panel presentations covered the following topics:
1. “Underwriting and Ratemaking for Catastrophe Risk”
Moderator: Robert B. Downer
Vice President and Actuary
Farmers Insurance Group
Panelists:
Jonathan White
Assistant Vice President and Actuary
Insurance Services Office. Inc.
Michael A. Walters
Consulting Actuary
MINUTES
OF THE 1995 SPRING MEETING
13
2.
“Driving on the Information Superhighway: Actuaries, Insurance and the Internet”
Panelists: Richard A. Derrig
Senior Vice President
Automobile Insurers Bureau of Massachusetts
James R. Garven, Ph.D.
Assistant Professor
University of Texas Finance Department
3.
“A World With GATT”
Moderator: John C. Narvell
Senior Consulting Actuary
Ernst & Young LLP
Panelists: Carl A. Modecki
President
National Association of Insurance Brokers
Kevin C.W. Mulvey
Associate Director, Government Affairs
American International Group
4.
“Commercial Package Ratemaking”
Moderator: Kathleen A. McMonigle
Director, Commercial Actuarial
ITT/Hartford
Panelists: Robert P. Eramo
Vice President and Chief Actuary
Johnson & Higgins
Gary Hoover
Actuary I
State Farm Insurance Company
5.
“Political Forces and Insurance Costs”
Moderator: Ronald J. Swanstrom
Principal
Coopers & Lybrand, L.L.P.
MINUTES OFTHE
14
1995 SPRING MEETING
Steven Ahmuty
Partner
Schaub, Ahmuty & Citrin
David Helfrey
Principal
Helfrey, Simon & James, P.C.
“Quality Assurance for the Actuarial Work Product”
Moderator: Robert F. Conger
Consulting Actuary
Panelists:
6.
Panelists:
7.
Thomas S. Carpenter
Arbella Mutual insurance Company
Roy G. Shrum
United States Fidelity and Guaranty Company
Jane C. Taylor
Reliance Insurance Companies
“The Impact of Reinsurance on Loss Reserves”
Moderator: Kim E. Piersol
Vice President and Actuaq
CNA Insurance Companies
Panelists:
Christy H. Gunn
Assistant Vice President and Associate Actuary
CNA Insurance Companies
Donald P. Skordenis
Senior Consultant
Alfred 0. Weller
President
Workers’ Compensation Reinsurance Rureau
MIKL’TES OFTHE
8.
1995 SPRING S1fXTING
“Current Financial Reporting Issues”
Moderator: Patrick J. Grannan
Consulting Actuary
Milliman & Robertson, Inc.
Panelists: Sheldon Rosenberg
Continental Insurance
Richard Roth, Jr.
Assistant Commissioner
California Department of Insurance
“Compilation of State Laws Regarding Statements of Actuarial Opinion on Loss Reserves”
Moderator: Robert J. Finger
Consulting Actuary
Panelists: Elise C. Liebers
Supervising Actuary
New York State Insurance Department
Michael L. Toothman
Consulting Actuary
Arthur Andersen LLP
10. “Actuarial Talent: Supply versus Demand”
Moderator/ Stephen P. D’Arcy
Professor, Department of Finance
Panelist:
University of Illinois
W.
James MacGinnitie
Panelists:
National Director of Actuarial Services
Ernst & Young LLP
David J. Oakden
Consulting Actuary
9.
16
MINUTES OFTHE
I’M
SPRING MEElING
1 1. “CAS Examination Committee”
Moderator: Curtis Gary Dean
American States Insurance Companies
Panelists: Michele A. Lombardo
Examinations and Information Systems
Administrator
David L. Menning
Senior Associate Actuary
State Farm Mutual Automobile Insurance
Company
12. “Questions and Answers with the CAS Board of Directors”
Moderator: Albert J. Beer
American Re-Insurance Company
Panelists:
Patrick J. Grannan
Consulting Actuary
John M. Purple
Consulting Actuary
Arthur Andersen, LLP
Richard H. Snader
Vice President, Corporate Actuary
United Stated Fidelity and Guaranty
Company
13. “CAS Actuarial Research Corner”
Moderator: Richard G. Woll
Senior Actuary
Allstate Research and Planning Center
An officers’ reception for new Fellows and guests was held
from 5:30 p.m. to 6:30 p.m., and the general reception for all
members and their guests was held from 6:30 p.m. to 7:30 p.m.
MINUTESOFTHE
1995SPRlNGMEETING
17
Tuesday, May 16, I995
Two general sessions were held from 8:30 a.m. to 10:00 a.m.
The first was
“Unfair Discrimination in Insurance Underwriting-Fact
or Fiction?’
Moderator: Irene K. Bass
Managing Director
William M. Mercer, Inc.
Panelists: Robert W. Klein, Ph.D.
Director of Research
National Association of Insurance
Commissioners
Eric F. Gottheim
The Robert Plan Corporation
Yvonne S. Sparks
Executive Director
Neighborhood Housing Services of
St. Louis, Inc.
The other session, presented simultaneously, was
“Technology and Information System”
Moderator: Stephen W. Philbrick
Consulting Actuary
Panelists: David Hollander
Partner
Andersen Consulting
John J. Kollar
Vice President
18
\lliiCTES
OFTHE 149 SPRIVG \IEETIS‘G
Stephen Jacobs
Partner
Reinhart. Boerner. Van Deuren. Norris. and
Reiselbach. S.C.
After a refreshment break, the concurrent sessions uere held
from lo:30 a.m. to noon.
Various CAS committees met from I:00 p.m. to 5:OO p.m. In
addition, concurrent sessions v,‘ere held from I:30 p.m. to 3:00
p.m.
All members and guests enjoyed a buffet dinner at the St. Louis
Zoo from 6:30 p.m. to 9:30 p.m.
Wednesday May 17, 1995
Concurrent sessions were held from 8:30 a.m. to IO:00 a.m.
Following the concurrent sessions. Jay Anpoff. Commissioner
of Insurance for the State of Missouri, gave a special presentation
to members.
From 10: 15 a.m. to 11:45 a.m.. a general session on Professionalism was held. This general session proifided case studies and
dramatizations of situations involving professional ethical quandries actuaries may face.
David Skurnick
Writer:
Senior Vice President and Actuary
F & G Re, Inc.
Nolan E. Asch
Director:
SCOR U.S. Corporation
Lauren Bloom
Cast:
General Counsel
American Academy of Actuaries
Jerome A. Scheibl
Member
Actuarial Board for Counseling and Discipline
19
Margaret W. Tiller
President
Tiller Consulting Group
CAS President Allan M. Kaufman officially adjourned the
1995 CAS Spring Meeting at noon after closing remarks and an
announcement of future CAS meetings.
May 1995 Attendees
The 1995 CAS Spring Meeting was attended by 162 Fellows,
130 Associates, and 89 Guests. The names of the Fellows and Associates in attendance follow:
FELLOWS
Manuel Almagro, Jr.
Richard R. Anderson
William R. VanArk
Nolan E. Asch
Irene K. Bass
Albert J. Beer
Phillip N. Ben-Zvi
JamesE. Biller
Ralph S. BlanchardIII
Ronald L. Bornhuetter
Paul Braithwaite
Randall E. Brubaker
CharlesA. Bryan
JamesE. Buck
Thomas S. Carpenter
SandersB. Cathcart
Mark M. Cis
Michael A. Coca
Robert F. Conger
Charles F. Cook
Timothy J. Cremin
Robert J. Curry
StephenP. D’Arcy
Ronald A. Dahlquist
Curtis Gary Dean
Michael C. Dolan
Robert B. Downer
Michael C. Dubin
Judith E. Dukatz
Grover M. Edie
ValereM. Egnasko
WarrenS. Ehrlich
DouglasD. Eland
Sholom Feldblum
GeorgeFescos
RobertJ. Finger
Nancy G. Flannery
JamesE. Fletcher
Michael Fusco
Scott F. Galiardo
Alice H. Gannon
RobertW. Gardner
Richard Gauthier
Owen M. Gleeson
StevenF. Goldberg
JamesF. Golz
Karen PachynGorvett
Linda M. Goss
Eric F. Gottheim
Leon R. Gottlieb
Bradley A. Granger
Patrick J. Grannan
Christy H. Gunn
David N. Hafling
JamesA. Hall III
Robert C. Hallstrom
David G. Hartman
Roger M. Hayne
Todd J. Hess
Anthony D. Hill
KathleenA. Hinds
JamesG. Inkrott
Richard M. Jaeger
RonaldW. Jean
Robert S. Kaplan
Allan M. Kaufman
20
MINUTES
Anne E. Kelly
Frederick 0. Kist
Joel M. Kleinman
Craig W. Kliethermes
John JosephKollar
Gary I. Koupf
Ronald T. Kozlowski
Israel Krakowski
GustaveA. Krause
Rodney E. Kreps
Andrew E. Kudera
Paul E. Lacko
Mathieu Lamy
Peter M. Licht
Elise C. Liebers
Robert F. Lowe
StephenJ. Ludwig
Aileen C. Lyle
W. JamesMacGinnitie
William G. Main
Stuart B. Mathewson
Kathleen A.
McMonigle
David L. Menning
Paul A. Mestelle
Robert J. Meyer
Glenn G. Meyers
Robert S. Miccolis
Brett E. Miller
Michael J. !vlillrr
David L. ~lillsr
Brian C. Moore
Phillip S. YInore
Nancy Diane Xlueller
OF THE 1995 SPRING MEETING
Donna S. Munt
JamesJ. Muza
Nancy R. Myers
John C. Narvell
Glen C. Nyce
TerrenceM. O’Brien
Paul G. O’Connell
David J. Oakden
Jennifer J. Palo
Curtis M. Parker
StevenC. Peck
StephenW. Philbrick
Kim E. Piersol
JosephJ. Pratt
EduardJ. Pulkstenis
John M. Purple
John F. Rathgeber
William P.Roland
SheldonRosenberg
Richard J. Roth, Jr.
David M. Savage
JeromeA. Scheibl
Brian E. Scott
Jeffery J. Scott
EdwardC. Shoop
Roy G. Shrum
David Skurnick
Oakley (Lee) E.
Van Slyke
Richard H. Snader
David Spiegler
Lee R. Steeneck
Russell Steingiser
JamesP. Streff
Stuart B. Suchoff
Christian Svendsgaard
Ronald J. Swanstrom
SusanT. Szkoda
JaneC. Taylor
CatherineHarwood
Taylor
Michael T. S. Teng
PatriciaA. Teufel
Kevin B. Thompson
MargaretWilkinson
Tiller
ThomasC. Tote
Michael L. Toothman
Gail E. Tverberg
JeanVaillancourt
John V. Van de Water
William Vasek
Gary G. Venter
Michael A. Walters
Bryan C. Ware
ThomasV. WarthenIII
Walter C. Wright III
Alfred 0. Weller
JonathanWhite
Charles Scott White
PeterG. Wick
Kevin 1,.Wick
JamesC. Wilson
Ernest I. Wilson
SusanK. Woerner
Richard G. Woll
21
ASSOCIATES
Rimma Abian
Martin S. Arnold
Allan A. Bell
Bruce J Bergeron
StevenL. Berman
Carol A. Blomstrom
Erik R. Bouvin
Douglas J. Bradac
Kevin M. Brady
JamesL. Bresnahan
Lisa J. Brubaker
Elliot R. Burn
J’ne E. Byckovski
SandraL. Cagley
PamelaJ. Cagney
Douglas A. Carlone
Jill C. Cecchini
Heather L. Chalfant
Jean-Francois
Chalifoux
PeggyCheng
Gary C. K. Cheung
Brian A. Clancy
Christopher J. Claus
William F. Costa
Christopher G. Cunniff
Laura B. Deterding
SeanR. Devlin
Behram M. Dinshaw
Pierre Drolet
StephenC. Dugan
Tammy L. Dye
Robert P.Eramo
S. Anders Ericson
JamesG. Evans
Gregg Evans
Andre F. Fontaine
SusanT. Garnier
ChristopherH. Geering
Eric J. Gesick
John T. Gleba
StevenA. Green
John E. Green
CharlesR. Grilliot
Leigh JosephHalliwell
Julie K. Halper
Timothy J. Hansen
David S. Harris
Philip E. Heckman
JosephA. Herbers
Betty-Jo Hill
John V. Hinton
JasonN. Hoffman
Brian L. Ingle
Christian Jobidon
Daniel K. Johnson
Daniel J. Johnston
Gail E. Kappeler
Lowell J. Keith
ThomasP. Kenia
Jean-Raymond
Kingsley
Brian S. Krick
Chung-Kuo Kuo
Marc LaPalme
Gregory D. Larcher
Debra K. Larcher
SteveE. Lehecka
Daniel E. Lents
Richard B. Lord
Cornwell H. Mah
Anthony L. Manzitto
Eduardo P.Marchena
SharonL. Markowski
Scott A. Martin
Tracey L. Matthew
Camley A. Mazloom
Deborah L. McCrary
Michael K. McCutchan
Kelly S. McKeethan
VanA. McNeal
Claus S. Metzner
Anne Hoban Moore
Kenneth B. Morgan, Jr.
Kevin T. Murphy
Hiep T. Nguyen
JamesL. Nutting
Dale F. Ogden
Milary N. Olson
ThomasPassante
Nicholas H. Pastor
ClaudePenlandIV
Robert C. Phifer
GenevievePineau
Robert E. Quane III
PeterS. Rauner
JamesE. Rech
Natalie J. Rekittke
Scott Reynolds
Meredith G.
Richardson
John W. Rollins
David A. Rosenzweig
22
John P.Ryan
SandraC. Santomenno
Christina L. Scannell
David 0. Schlenke
Michael J. Scholl
Robert D. Share
Jeffrey S. Sirkin
Donald P. Skrodenis
M. Kate Smith
Byron W. Smith
EugeneG. Thompson
John B. Sopkowicz
Eric Vaith
Klayton N. Southwoc
3d Cynthia L. Vidal
Michael J. Sperduto
David G. Walker
StevenJ. Symon
RobertJ. Walling III
Joy Y. Takahashi
Michael W. Whatley
Trina C. Terne
StevenB. White
David M. Terne
Michael J. Williams
William F. Wilson
Daniel A. Tess
Volume LrcyXII,
Part 2
No. IS7
PROCEEDINGS
November 12,13,14,15,1995
BALANCING
TRANSACTION
COSTS AND RISK LOAD
IN RISK SHARING ARRANGEMENTS
CLIVE
L. KEATINGE
Abstract
In formulating
eficient risk sharing arrangements, it
is desirable to minimize both transaction costs and the
risk load required by the participating
insurers. A simple
yet realistic model that explicitly incorporates both transaction costs and risk load is put forth in this paper It is
shown that, under very general conditions, the optimal
risk sharing arrangement which results is constructed in
layers. Remarkably simple expressions are given for the
optimal boundaries between layers as well as each participating insurers share of each layer: Several examples
are included that illustrate the application of the model.
1.
INTRODUCTION
This paper addresses the related subjects of optimal risk sharing and premium calculation. “Risk sharing” refers to an ar23
24
BALANCING
‘I’KANSA(“llON
(‘OS’I’S ANI)
RISK I.OAi)
rangement among various entities (in an insurance context. usually insureds, insurers, and reinsurers) to share in the payment
of losses. “Premium calculation” refers to the process of figuring charges to add to expected losses to obtain premiums for a
particular risk sharing arrangement. These charges take into account both the transaction costs (e.g., commission, brokerage,
and overhead) and the risk load associated with a risk sharing arrangement. Optimal risk sharing and premium calculation
have been discussed quite frequently in the actuarial literature.
The primary feature of this paper that distinguishes it from most
other treatments of these subjects is the explicit inclusion of
transaction costs as an integral part of the model used to derive
results.
The problem discussed in this paper is that of finding the
risk sharing arrangement that minimizes the combined premium
charged by all of the insurers sharing a particular risk.’ In the
model used to address this problem, we assume that each insurer charges a specified percentage of its own expected losses
to account for transaction costs and a specified percentage of
the variance of its own losses to account for risk load. These
percentages may differ by insurer. In general, we expect insurers that tend to take on small amounts of expected losses for
each risk (often reinsurers) to have transaction costs that are a
larger percentage of their expected losses than insurers that tend
to take on large amounts of expected losses for each risk. Likewise, we expect insurers that tend to take on a very large number
of risks (often reinsurers) to have risk loads that are a smaller
percentage of the variance of their losses than insurers that take
on a small number of risks. More will be said about this later.
‘For convcnicncc, throughout this paper, the term “insurer” will bc uxd to rcfcr to any
participant in a risk sharing arrangement. Howcvcr, all the participants in a risk sharing
arrangement need not bc insurers. An insured may retain a portion of its own losses, a
rcinsurcr may assume losses through a primary insurer, or the risk sharing could bc in
a noninsurance context. In the cast of an insured retaining a portion of its own losses,
although prcmlum would not change hands, the msurcd would incur a cost in maintaining
the additional capital and liquidity ncccssary to ahsorb the retained Iosscs.
BALANCING
TRANSACTION
COSTS AND
RISK LOAD
25
There are certainly other ways one could account for transaction costs in a risk sharing arrangement, and risk load has been
a subject of ongoing debate for many years. The purpose of this
paper is not to debate the merits of various methods of handling
transaction costs and risk load. The model described in this paper is useful because it is simple enough to yield results that are
mathematically tractable yet realistic enough to yield results that
provide real insight.
2. THE PROBLEM
We begin with the usual formulation of the collective risk
model. Let N denote the number of claims produced by a risk
(or portfolio of risks) in a given time period. Let Xi, X2, X3,. . .
denote the various claim sizes. We assume N, X1, X2, X3,. . . to
be mutually independent random variables and X1, X2, X3,. . . to
be identically distributed. If S = X1 + X2 + ... + X,, then:
E[S] = E[N] . E[X],
and
Var[S] = E[N] . Var[X] + Var[N] . (E[X])2
= E[N] . E[X2] - (E[X])2 + 3.
=E[N].{E[X2]+
(s-1)
.(E[X])‘}.
Next, we assume that there are C insurers available to share
in the payment of losses. Further, we assume that each insurer
pays a predetermined percentage of each claim. These percentages may vary by claim size. Thus, each insurer has associated
with it a payment function, p(x), which can vary between 0 and
1, that indicates the percentage of each claim that the insurer will
pay.2 If Si designates the total losses paid by the ith insurer and
*Payment may also be based on the sum of all claims arising from each occurrence. For
convenience, the term “claim” will be used throughout this paper.
26
BALANCING
TRANSACTION
COSTS AND RISK LOAD
pi(x) is the payment function for the ith insurer, then:
E[Si] = E[N] . E[pi(X) oX],
and
Var[Si] = E[N] o{E[(pi(X) . X)2]
+
Var[N] _ ~
) . (E[pi(X).
E[N]
(
X])2}.
Let @ibe the percentage of its own expected losses charged by
the ith insurer to account for transaction costs, and let vi be the
percentage of the variance of its own losses charged by the ith
insurer to account for risk load.3 Then the combined premium
charged by all of the insurers sharing the risk is:
c
M = ~(E[5’i] + @ioE[Si] + vi . Var[Si])
i=l
c
=
E[S] + ~(@i . E[Si] + vi . Var[Si]).
i=l
The problem is to find the payment functions for each of the C
insurers that minimize M subject to the constraints that:
c
0< pi(X)
<1
and
~pi(x)
i=l
= 1.
3. THE SOLUTION
The solution is given here without proof. The proof is provided in the Appendix. First, we assume that the C insurers have
been arranged so that the following relation holds:
The solution then involves the familiar concept of layering. The
optimal risk sharing arrangement is organized into C layers.
3@i is dimen~i~~~s
convenience,
$–’
and vi
hm dimension
will be omitted
throughout
$– 1 (if we are working
this paper.
with dollars).
Rr
BALANCING
TRANSACTION
COSTS AND RISK LOAD
27
The first layer (from zero to the first layer boundary) is allocated entirely to the first insurer.
The second layer is allocated to the first two insurers in the
following proportions:
Insurer 1:
l/7#~
l/1#~ + l/?/J~’
Insurer 2:
l/$!l~
l/~~ + l/+~”
and
Thus, for a claim that penetrates the second (but
layer, the first insurer pays the entire portion of
falls below the first layer boundary and a fraction
above it. The second insurer pays a fraction of the
not the third)
the claim that
of the portion
portion
of the
insurers
in the
claim above the first layer boundary.
The third layer is allocated
following proportions:
Insurer 1:
Insurer
2:
Insurer
3:
to the first three
1/+,
l/?/J~ + l/+~ + l/?/)~’
l/~,
1/+2
+ l/7)~ + l/+q’
and
l/q!q
1/41 + 1/42 + 1/+3”
One insurer is then added in each successive layer until the
top layer has all of the C insurers participating in the following
proportions:
Insurer
i:
l/~i
1/+1 +1/7/J~ +..+
l/?#c”
Thus, for low layers, which contribute much more to expected
losses than to variance, only the insurers with the smallest #is
participate.
For high layers, where variance is a much more important consideration
than expected losses, many insurers partic-
28
BALANCING
TRANSACTlON
COSTS AND
RISK LOAD
ipate in order to better reduce the variance. Within a particular
layer, the insurers with the smallest ?,!J~s
get the largest shares.4
We now address the issue of the location of the layer boundaries. Let the layer boundaries be I1 5 125
< 1~~1. Then each
lj is given by the solution of the following equation:
I, + --l
Var[Nl
I (
VW
1
‘E[X;lj]-1
’
i=l
@j+l
- @i = ()
2’tii
where E[X;f;] is the expected value of X limited at I,.’ The
relationship between adjacent ljs can be expressed as follows:
(lj
-
/j-l)
+
N]
~Var[
E[N] - 1 . (E[X;lj]
- E[X;lj-11)
The first thing to observe about these equations is that the fjs
depend on the claim count distribution only through the ratio
of the variance to the mean. If this ratio is 1, as it is with the
Poisson distribution, the ljs are independent of the claim size
distribution. If this ratio is less than 1, the more severe the claim
size distribution, the higher the fjs will be. If this ratio is greater
than 1, the more severe the claim size distribution, the lower the
fjs will be.
The second of the above equations shows that if $j is associated with the insurer just added in a given layer and $j+l is
‘This same type of
context of a model
5This equation can
tive of the left side
layering arrangement was derived by Buhlmann and Jcwcll [l] in the
based on exponential utility functions.
be easily solved for I, using Newton’s method. Note that the derivaof the equation with respect to I, is simply:
WNI
I + --1
( EINI
where F(x)
is the cumulative distribution
)
-(l--F(l)))
function of A’.
BALANCING
TRANSACTION
COSTS AND
29
RISK LOAD
associated with the insurer to be added in the layer above, then
the greater the difference between them, the greater the width
of the former layer will be. In other words, if the insurer to be
added in the layer above charges a much greater percentage of
its expected losses than the most expensive of the insurers participating on a given layer, a large increment will be required to
reach a point where the reduction in variance provided by the
addition of the next insurer is worthwhile. On the other hand, if
$j = $j+i, then the width of the layer will be 0, and both insurers will be added at the same time. In the extreme case where
all of the @is are equal to one another, all of the Ijs will be
0, so there effectively will be only one layer, with all C of the
insurers participating. This reflects the well-known result that if
transaction costs do not depend on how a risk is shared, then a
quota share arrangement is optimal.
Another noteworthy aspect of the above equations is that
a given lj is only affected by the $J~Sassociated with insurers
on layers below it. Thus, the +is associated with insurers to be
added in higher layers have no effect on the location of a particular fj. It is also clear that smaller qis will result in higher fjs. In
other words, if insurers do not charge large percentages of the
variances of their losses, variance reduction is less of a priority
than it would otherwise be, and the points at which insurers are
added can be higher.
The optimal risk sharing arrangement described above minimizes the combined premium charged by all of the insurers sharing a risk. In order to calculate each insurer’s premium, expressions are needed for E[pi(X) . X] and E[(pi(X). X)‘], the first
and second moments of each insurer’s own claim payment distribution. Since the optimal risk sharing arrangement is constructed
in layers, the needed expressions are as follows:
E[Pi(X)
* x] = f: ‘ij l” (1 - F(X))dX,
j=i
I 1
and
30
tlAl.AN(‘IN(i
‘I‘RANSA(‘XON
(‘OS’I S ANI)
RISK I.OAl)
2
+
rij
.
(X
-
ij
1)
.f(
.U)dX.
,
i
where:
fo = 0 and IC = 00,
I(X) = probability
density function of A’, and
rlj = ith insurer’s share of the jth layer as defined above
If claims are censored by a policy limit, a term must bc added
to the expression for the second moment to take into account
the spike of probability at the policy limit. However, the equation used to calculate the lIs is not affected by a policy limit. Insurcrs that participate only on layers that fall completely above a
policy limit are effectively not needed in the optimal risk sharing
arrangement.
4.
EXAMPI.liS
The application of the results presented in the previous section will be illustrated with several examples. The claim size distributions used in the examples are Mixed I’aretos.’ Each distribution is the weighted average of two Paretos. one of which has
a relatively thick tail and one of which has a relatively thin tail.
The density and distribution functions of the Mixed Pareto are
as follows:
IViI.AN(‘ING
Il~ANSA(“IlON
(‘OSIS
ANI)
31
RISK I.OA[)
TABLE 1
KEY STATISTICSFOR OPTIMAL RISK SHARING
TYPICAL GENERAL LIABILITY RISK
BI = 25.OM)
B-3 = 5,000
QI = 1.25
Q2 = 3.2s
P = .x0
Var[N]/E[A’]
= 2
Layer
Layer
Layer
Layer
1 Share
2 Share
3 Share
4 Share
Expcctcd Loss
Q, Charge
I/J Chap
Total Charge
Pcrccntagc
Lavcr
Layer
Layer
Layer
I:
2:
3:
4:
0 - 75,677
75,677 - 255.814
255,814 - 562,307
562,307 - 1,0(X),000
Polq
Limit = 1,oOO,OoO
lnsurcr
1
Insurer
2
lnsurcr
3
lnsurcr
4
Total
100.0%
45.5
27.0
17.5
54.5%
32.4
21.1
26.3
3.5.1s
100.0
2,.5x9
1,057
414
13,874
No Risk
Sharing
13,874
259
107
366
1%
38
1%
83
x
91
991
468
1,459
694
1,401
2.09s
18.5%~
22.0’%!
lO..5’Z
15.1%
14.1%)
The charges in the examples are calculated assuming that
E[N] = 1. Charges for other values of E[N] can be found simply
by multiplying the charges shown by E[N].
In each of the examples, we will assume that there are six
insurers available to share the risk. The @is for the six insurers
are 45, .lO, .15, .20, .25, and .30, and the ?+!J;s
for the six insurers
are .30. 10P6, .25- 10P6, .20. lo-‘, .15. l@, .lO. lo-“, and 45.
1V6, respectively. More will be said later about how these values
might be estimated.
Table 1 shows the key statistics for the optimal risk sharing arrangement for what might be considered a typical general liability risk with a $1,000,000 policy limit. Note that only four insurers
are required in this case. For comparative purposes, charges are
also shown for the case in which there is no risk sharing and In-
32
HAI.AN(‘ING
.I-RANSA(“llON
(‘OSI’S AND RISK 1,OAl)
TABLE 2
LARGE POLICY
BZ = 25,(xX)
B2 = S,oOO
(jr = 1.25
Q-7 = 3.25
P = x0
Var[NJ/EIN]
Laycr
1 share
L.aycr
Layer
Layer
Layer
Layer
2
3
4
5
6
Share
Share
Share
Share
Share
Q, Charge
1c,Charge
Total Charge
Percentage
Layer 1:
Layer 2:
Layer 3:
Layer 4:
L..aycr 5:
= 2
Layer 6:
Insurer Insurer
1
2
100.0%
45,s
27.0
17.5
11.5
6.8
0 - 75,677
75,677 - 255,814
255,814 562,307
562,307 1,036,0.58
1,036,(158~ 1.760,102
1,7M),102 ~~lO,oOO,Oo()
Insurer Insurer Insurer
3
4
5
Poliq
lnsurcr
6
Tol~Il
100.0%
54.5%
32.4
21.1
13.8
x.2
40.6%
26.3
17.2
10.2
3.5.19
23.0
13.6
.34..5%
20.4
40.8%
10,099
2,932
1,485
YX6
822
Y84
SOS
401
906
293
193
486
223
124
347
107
Y4
291
206
86
292
205
109
304
16.6%
23.4%’
29.5%
35.5%
4 1.1%
9.0%
l.imit = 10,000,000
100.0
100.0
loo.0
100.0
1MM
No Risk
Sharing
17,308 17,308
1,719
1,007
2,726
15.7%
866
X,722
9,588
55.4%
surer 1 takes the entire risk. Risk sharing in this example results
in a savings of 4.6% of expected losses.
Table 2 shows the statistics for the optimal risk sharing arrangement for a risk identical to that underlying Table 1 except
with a policy limit of $10,000,000. In this case. all six insurers
are required, and risk sharing results in a savings of 39.7% of
expected losses.7
‘This 1s an illustratmn of how the abscncc of risk sharing m a model can result in very
large risk loads at high poliLy limits. Robbin [2] has discussed the need to consider risk
sharmg when computing risk loads and has prcscntcd a simple model of risk sharing
(alkwing only quota share arrangcmcnts) that incorporates transaction costs (attributed
IO Klmkcr).
BALANCING
TRANSACTION
COSTS AND
33
RISK LOAD
TABLE 3
VARIANCEEQUALS MEAN
Bl = 25,000
82 = 5,000
Ql = 1.25
Q2 = 3.25
P = .&I
Var(N]/E[N]
Layer
Layer
Layer
Layer
1 Share
2 Share
3 Share
4 Share
Expected Loss
9 Charge
$ Charge
Total Charge
Percentage
Layer
Layer
Layer
Layer
1:
2:
3:
4:
0 - 83,333
83,333 - 266,667
266,667 - 575,000
575,OOU- 1,000,000
Policy Limit = 1,000,000
= 1
Insurer
1
Insurer
2
100.0%
45.5
27.0
17.5
54.5%
32.4
21.1
9,978
Insurer
3
Insurer
4
Total
40.6%
26.3
3S.lF
1OO.O?
100.0
100.0
100.0
2,471
1,022
397
13,874
No Risk
Sharing
13,874
499
301
800
248
103
351
153
36
189
80
8
88
980
448
1,428
694
1,343
2,037
8.0%
14.2%
18.5%
10.3%
14.7%
22.2%
Table 3 shows the statistics for the optimal risk sharing arrangement for a risk identical to that underlying Table 1 except
with Var[N]/E[N]
equal to 1. As noted in the previous section,
in this case, the layer boundaries are independent of the claim
size distribution. The results are similar to those shown in Table 1.
Table 4 shows the statistics for the optimal risk sharing
arrangement for a risk identical to that underlying Table 1 except with smaller QI and Q2 parameters. This adjustment
thickens the tail of the claim size distribution, thus making
risk sharing more important. The layer boundaries change very
little, but risk sharing results in a savings of 9.7% of expected
losses.
34
BALANCING
‘~RANSAC’I‘ION
(‘OS73
ANI)
KISK l.O/\I~
TABLE 4
SMALLER Ql AND Q2 PARAMETERS
Bl =25,000
B2=5,000
Ql = 0.75
Q2 = 2.15
P =.80
Var[N]/E[N]
Layer
Layer
Layer
Layer
Layer
Layer
Layer
Layer
= 2
1:
2:
3:
4:
0 - 72,930
72,930 - 248,026
248,026- 548,936
548,936 - 1,OW,OOO
Po!iq Llml! = 1.~,KlO.OoO
Insurer
1
Insurer
2
lnsurcr
3
Insurei
4
Total
17.352
8,338
4X)?
?.iYY
3_7.544
868
910
1,778
8.74
450
1,284
‘21
10
Y20
470
7,YOZ
l.hl3
.1,‘;15
15.4”;
19.1’;
‘2.2c;
1 Share
2 Share
3 Share
4 Share
Expcctcd Loss
4 Charge
+ Charge
Total Charge
Percentage
10.2’;
54
533
ii.“;
is91
1.645
6,039
7.68‘1
3.45
5. AGGREGATION
To this point, we have assumed that risk sharing is done on a
claim by claim (or occurrence by occurrence) basis. Each insurer
participating on a risk pays a predetermined percentage of each
claim, with the percentage depending on the size of the claim.
However, the model can also be applied to situations where a
number of claims are aggregated together before being allocated
to each insurer. If the claims are independent of one another,
algorithms are available that may be used to calculate an aggregate distribution from the underlying claim count and claim size
distributions, or a simulation technique may be used.
The only change to the model involves the equation for the
layer boundaries. If claims are aggregated together over definite
time periods, there will be only one “claim” per time period.
Therefore, the variance-to-mean ratio of the claim count distribution must be set at zero, and the equation reduces to:
lj -
j
E[X; fj] - C
- $i = o
@j+l
i=l
2’$i
’
An advantage of aggregating independent claims together before allocating them to insurers is that claims considered as a
group are more predictable than claims considered individually.
As a result, more of the expected losses can remain in the lower
layers with insurers with lower @is, thus resulting in a lower combined premium for each risk. The larger the number of claims
aggregated, the greater the effect will be.
A lower bound for the combined premium may be easily computed. First, note that the combined charge for transaction costs
cannot be lower than the total expected losses multiplied by &,
which we have assumed to be the smallest of the @is. Second, as
alluded to earlier, if transaction costs are disregarded, a quota
share arrangement is optimal, with each of the insurers being allocated relative shares inversely proportional to their @is. Therefore, a lower bound for the combined premium may be obtained
by assuming that all of the expected losses are allocated to the
lowest layer and all of the variance is allocated to the highest
layer. This lower bound is thus:
I\
EIS]+O1.EIS]+&
. Var[S]
i=l
= E[S] + 0, . E[S] + c
C
/=I
1
. Var[S].
l/Qi
Table 5 shows how this lower bound compares to results without risk sharing and with risk sharing on a claim by claim basis
for the risks underlying Tables 1 and 2.
36
BALANCING
TRANSACTTON
COSTS AND
RISK LOAD
TABLE5
TOTAL CHARGEASA PERCENTAGEOFEXPECTED LOSSES
Policy
Limit
No Risk
Sharing
Claim By
Claim
Aggregate
Lower Bound
1 ,ooo,m
15.1%
55.4
10.5%
15.7
5.7%
8.4
10,000,000
Given that it is possible to lower the combined premium
by aggregating claims together before allocating them to each
insurer, one might conclude this should always be done. However, this may not always be the best approach. For example, if
claims are aggregated together, an insurer participating on a high
layer can be affected by a large number of small claims in addition to one large claim, which may not be desirable. In some
cases, the overhead associated with aggregate coverage may
result in larger transaction costs. The mode! cannot account
for all the practical realities that must be considered. Also,
to reduce the combined premium by a significant amount, it
may be necessary to aggregate together a very large number of
claims.
Finally, it should be noted that many risk sharing contracts
exist that aggregate together only the portion of claims in specified layers. For example, in many retrospective rating contracts,
losses below a given retention are aggregated together before
determining coverage, while the insurer pays the portion of any
claim that falls above the retention. As another example, a reinsurer may provide coverage only if the sum of a!! losses that fall
in a given layer exceeds a given aggregate retention, while the
ceding insurer retains a!! losses in this layer below the aggregate retention as we!! as a!! claims that fall completely below the
layer. The hybrid nature of these contracts makes them difficult
to analyze. However, the expression giving a lower bound for the
combined premium remains valid.
BALANCING
6.
TRANSACTION
PARAMETER
COSTS AND
RISK I.OAI)
37
ESTIMATION
This section addresses several points that must be considered
if we wish to estimate actual values for the @isand +is. As shown
previously, each lj is given by the solution of the following equation:
Note that only differences of &s appear in this equation. If a!!
the $jS were increased by the same amount, the solution to the
equation would not change. This reflects the fact that what matters are differences in transaction costs among insurers. If there
are some costs (e.g., agents’ commissions) that are incurred regardless of how a risk is ultimately shared among insurers, then
these costs have no effect on the optima! risk sharing arrangement.
If risk sharing is accomplished through reinsurance, the difference between $i for a primary insurer and C#J~
for a reinsurer should reflect the additional transaction costs (e.g., brokerage and overhead) that are incurred as a result of the reinsurante contract. Reinsurers that take on small amounts of expected
losses for each risk, such as those that tend to take on high layers, can be expected to have larger @is than reinsurers that take
on large amounts of expected losses for each risk, such as those
that tend to take on low layers.
The estimation of JLJ~
for an insurer should generally be somehow based on the variability of the insurer’s overall results. A
simple estimation method is illustrated here. Suppose an insurer
estimates that its aggregate loss distribution for the next year (for
losses retained) has a mean of $50,000,000 and a standard deviation of $5,000,000 (and thus a variance of 25,000,000,000,000 $2).
Suppose further that the insurer decides that it needs half of the
standard deviation, or $2,500,000, as risk load. Then, in order to
generate the required amount of risk load, its ?#j should be cal-
38
13AIANCING
TRANSA(‘TlON
(‘OSI’S AN11 RISK l.OAI)
culated as follows:
$i
=
2,500,OOO
= .lO’ 1oP.
25,000,000,000,000
Since variance is additive for independent risks (or independent
blocks of risks), if the insurer uses this $i when calculating the
risk load for each of its independent risks (or independent blocks
of risks), the required amount of risk load will be generated.
It may be difficult to obtain estimates of $; and @i for each
insurer participating in a risk sharing arrangement. However, if
a primary insurer is simply interested in finding the retention below which it should retain 100% of every risk, and if the insurer
is willing to assume that Var[N]/E[N]
is 1, then the equation at
the beginning of this section simplifies to:
Thus, the insurer needs only an estimate of the additional transaction costs associated with the most inexpensive acceptable
reinsurance available and an estimate of its own ~9,. For example, if & - C#J~
is estimated to be 45 and +, is estimated to be
.lO. 10P6, then:
.05
= $250,000.
11 =
2 * .lO. 10-h
The final topic of this section is the effect of trend in claim
sizes, When a trend factor T is applied to a claim size distribution. we expect the optimal layer boundaries to be multiplied
by T. If we examine the equation at the beginning of this section we see that this will occur if a!! of the @is are divided by
T. Since the variance of each insurer’s losses is multiplied by T‘when T is applied to a claim size distribution, each insurer’s risk
T' = T. Since each insurer’s
load would be multiplied by (l/T).
expected losses would also be multiplied by T, each insurer’s risk
load as a percentage of expected losses would remain constant.
If nothing else changes. this is the desired result. Thus. whenever
a claim size distribution is trended. the d’is must be “detrended.”
The @is are not affected.
HAI.AN(‘IN(i
7.
IH,WSA(‘[-ION
THE
(‘OS I.5 AYI)
REINSURAXCE
III%
I OA1)
hMRKET
This section is a brief discussion of a few issues that relate to
how the mode! and its results fit into the actual workings of the
reinsurance market. within which most risk sharing among insurers takes place. First, to this point, no mention has been made
of allocated loss adjustment expenses (ALAE). If ALAI? is included with losses before being allocated to layer, ALAE may
be incorporated into the model by using a claim size distribution that is based on the sum of losses and ALAE. If ALAE is
allocated to layer in the same proportions as the losses, ALAE is
not easily incorporated into the mode!. However, setting aside
any practical considerations, this treatment of ALAE is less efficient from a risk sharing perspective than including ALAE with
losses. A clear illustration of this occurs when a ceding insurer
incurs a large amount of ALAE in defending a claim on which
ultimately no payment is made. In this case. risk sharing does
not occur; the ceding insurer pays the entire ALAE amount.
In the examples presented earlier, the insurers with the
larger @is, presumably reinsurers, were also assumed to have the
smaller +is. An examination of the mode! shows that this relationship does not necessarily have to hold. Although large reinsurers may indeed have small +is, there is also room for reinsurers with large $iS. They would simply receive smaller shares of
the layers on which they participate.
One apparent drawback of the mode! is that, in order to apply
it, we must assume that a set number of insurers are available
to participate in a risk sharing arrangement. In reality, numerous insurers and reinsurers may be competing to participate on
a particular risk. In the examples presented earlier, we assumed
that there was only one insurer with a $i of 45 available to participate in the risk sharing arrangement. In reality, there may be
numerous insurers with @is of .05 available to participate in the
risk sharing arrangement. If the mode! were strictly applied, a!!
of the insurers would participate, each receiving a relatively small
share of the expected losses. However, if this were to occur, it is
doubtful that the @is of these insurers would remain at .05. It
40
BALANCING TRANSACTTON (‘OSTS AND RISK I.OAD
is likely that their transaction costs as a percentage of their expected losses would increase.
This illustrates an implicit assumption underlying the mode!,
namely that each insurer’s $i is reasonable given the amount of
expected losses taken on by each insurer for a particular risk.
Too many insurers participating in a risk sharing arrangement
simply drives up the @is for a!! of them.* At some point, this
offsets the reduction in variance achieved by incorporating extra insurers on a risk. If a number of insurers with @s of .05
were to compete for a particular risk, in reality only one of them
would end up participating on the risk. The higher layers would
be left to the reinsurers, with larger @is, that specialize in taking
on small amounts of expected losses for each risk.
Thus, for purposes of finding the optima! risk sharing arrangement and its associated premiums, we can assume that a limited
number of insurers are available to participate. It would certainly
be possible to construct the $i for each insurer as a function of
the expected losses it takes on, instead of as a fixed value. However, the danger in doing this is that the mathematical complications introduced may obscure any additional insight that might
be achieved. The allure of the mode! as it stands is that it captures the essential features of the problem being addressed, yet
is still simple enough to yield a tractable solution.
8.
CONCLUSION
In formulating risk sharing arrangements, if transaction costs
are minimized without accounting for risk load, then the conclusion is that risk sharing should not take place. If risk load
is minimized without accounting for transaction costs, then the
conclusion is that every risk should be shared pro rata among as
many insurers as possible. Clearly, neither conclusion is correct.
The mode! described in this paper provides a workable way to
find the risk sharing arrangement that strikes the best balance
between the two competing goals of the minimization of transaction costs and the minimization of risk load.
8An alternate point of view is that additional insurers bring with them additional
costs instead of larger 4,s. Either way, the effect IS the same.
fuced
BALANCING
TRANSACTION
COSTS AND
RISK LOAD
41
REFERENCES
[l] Biihlmann,
ASTIN
H. and W. S. Jewell, “Optimal
Bulletin 10, 1979, pp. 243-262.
Risk Exchanges,”
[Z] Robbin, I., Discussion of Meyers: “The Competitive fiarket Equilibrium Risk Load Formula for Increased Limits
Ratemaking,” PCAS LXXIX, 1992, pp. 367-384.
42
MI.AN(‘ING
‘l’RANSA2(‘IlON
(‘MIS
i\VII
RISK I 0~1)
APPENDIX
1.
PRELIMINARIES
The problem addressed in this appendix is that of finding the
set of payment functions {P;(X)} for the C insurers that minimizes:
A4 = E[S] + c($,
E[(p,(X)
E[S,] + (itI Var[S,I)
X,‘]
+
Var[N]
,
-~
1-qN ]
(r:[p,(*Y).
q2
,
subject to the constraints that:
There are three basic steps to the proof of the solution. corresponding to the remaining three sections of this appendix. In
the first step we show that any set of payment functions minimizing M must satisfy the condition that an insurer which pays a
given amount on a claim of a given size pays at least as much
on claims of all larger sizes. This implies that the number of
insurers participating in the payment of a claim may not decrease (and may very well increase) as the size of the claim increases. In the second step, we use the method of Lagrange multipliers to find a condition that must hc satisfied by any set of
payment functions minimizing M given that the expected losses
allocated to each insurer are fixed at certain amounts. It can
then be deduced that the only risk sharing arrangement satisfying both these conditions is a layering arrangement with one
I3AIANCING
TRANSACTION
COSTS AND
RISK I.OAD
43
insurer added at each successively higher layer. Finally, in the
third step we find the layering arrangement minimizing M without restricting the amount of expected losses allocated to each
insurer.
Similar reasoning applies regardless of whether the claim size
distribution is discrete, continuous, or mixed. However, to make
the proof easy to follow, we use a discrete formulation in the
first two steps and a continuous formulation in the third step.
In the second step, we assume that claims may take on integral
values from 1 to 00 and that each possible value has positive
probability. In the third step, we assume that the claim size distribution has a probability density function that is positive everywhere. The assumption of positivity does not restrict the generality of the solution, because any probability or probability density
function that vanishes in some places can be approximated by a
function that is positive everywhere. yet where the contribution
to M from points or intervals that actually have zero probability is arbitrarily small. Thus, with the proviso that the payment
functions may take on arbitrary values where the probability
or probability density function of the claim size distribution is
zero, the solution holds for any claim size distribution with finite
mean and variance (which is necessary for the problem to make
sense).
2.
A FIRST NECESSARY
CONDITION
We will now show that if M is at a minimum and XL and
XR are any two possible claim sizes such that XL < SR, then
pi(XL)XL < pi(xR)xR for each of the C insurers. In other words,
any set of payment functions minimizing M must satisfy the
condition that an insurer which pays a given amount on a claim
of a given size pays at least as much on claims of all larger
sizes.
SUppOSe
pi(xR)xR
that for some XL and ,rR. XL < X’R and pi(X.L)XL >
for at least one insurer. Let one of these insurers have
BALANCING
44
TRANSACTION
COSTS AND RISK LOAD
index 1 and let D1 = pl(x~)x~ - pl(xR)n~. DiS associated with
the other insurers may then be selected such that the following
conditions are satisfied:
Di ~0,
Pi(XL)XL
- pi(xR)xR
2 Di
< 0,
D* +fjDi
and
if
PiCxL)XL
2 P~(XR)XR,
if
Pi(xL)xL
< pi(xR)xR,
=O.
i=2
Now let an alternate set of payment functions {P:(X)}
fined as follows:
m> =Pi(X)
- f(XR) .-Di
f@L)
PtCx)
= PiCx)
+
pi*(x) = Pi(X),
fh)
.-
+ f(xR)
Di
if
X=XR,
and
XR’
otherwise,
where f(x) is the probability
tion. Then:
PT(XL)XL~(XL)
x = XL,
XL ’
+ f(XR)
f(XL)
if
be de-
function of the claim size distribu-
+ PT(XR)XR~(XR)
f&d
f(XL)
+ ftXR)
xL
/(XL)
.
fcXL)
= Pi(XL)XLf(XL)
+ ftXR)
+ Pi(.rR)XRf(XR).
9
XR I
XRJ‘(XR)
BALANCING
Thus, E[pf(X)
TRANSACTION
. X] = E[pi(X).
COSTS AND
45
RISK LOAD
X]. Also:
(Pt(XL>XL>2f(XL> + (Pt(xR)xR)2f(xR)
f(XR)
= Pih) ([
f(XL)+ fcXR) xL
([
+
pi(xR) +
- 2Di[pi(xL)xL
f(xd
SW 1 >
+ fcxd
'-
Di
XR
xR
'f(xR)
- PI(XR)XR~. fcXLI + fcxRj
+ 0”. fcxd ’ fcXd
ftXL) + f(XR)’
Since DI = pl(x~)x~
-
pl(xR)xR:
(p;(x~)X~>~f(x~>
+ (PT(~R)XR)~~(XR>
= (PI(XL)XL>~~(XL) + (PI(XR)XR)~/(XR)
_ 0:.
f(‘L) * ftxR)
fcxL) + fcXR)
< (PI(XL>XL)~~(XL.> + (PI(XR)XR)~~(XR>.
For i # 1, since
pi(x~)x~
- pj(XR)XR
(pf(x~>x~>~f(x~>
_ 0”.
5
Di 5 0:
+ (PT(XR>XR)~S(XR>
f(XL)-f(XR)
f(XL) + ftxR)
<(pi(x~)x~)~f(x~)+(pi(x~)x~)~f(x~)~
Thus,
EKpf(W.
E[(p;(X)
XI21i
+ X)2] < E[(pI(X)
Wpi(X)
. W21.
. X)2]
and
for
i # 1,
46
~1AIANCIN(i ‘I’RANSA(“I-ION (‘OS1 S ANIl RISK l.OAD
Therefore, the alternate set of payment functions {p;(x)}
produces a smaller value of A4 than that produced by the original
set of payment functions. Hence, if M is at a minimum, pi(.~,+~
may not be greater than p;(-r~)x~ for any insurer.
3. A SECOND NECESSARY
CONDITION
We will now show that the optimal risk sharing arrangement
must be constructed in layers. with one insurer added at each
successively higher layer.
To ensure that 0 5 pi(x) 5 1, let pi(x) = $(.u). We will then
optimize each zi(x), which for notational convenience will be
written as simply zi. Also for notational convenience, let v =
Var[N]/E[N].
In the long expression for M. we will drop the
leading factor E[N] and the leading term E[X] in the brackets
since neither one will have an effect on the solution. Thus. we
are left with the problem of minimizing:
EL2; X] + ,I/:, { E[(z’
MI = x(4,
X)‘]
+ (L, - I). (L</z,2 x]y),
r=l
subject to the constraint that:
Now let Zi = E[$. X]. For now. we will assume that the Zis
are fixed. Later, we will find optimal values for the Z,s. Thus, we
want to minimize:
C
M,
=c
qJ,~Zjf~:,~
i=l
c
I &z,l.r-‘/(r))+(\v
I).Z”
L:
I,
= f-(-J;. z; + vj. (L,- ] ) z;, + f; >‘: l$;z;‘sq-(x)
i=l
i=l
t=l
tMI.AN(‘ING
‘IRANSA(“I‘ION
(‘OS’I’S AND
47
RISK l.OAII
subject to the constraints that:
z,?
=1
and for each i,
r=l
i=l
To find the z;s minimizing M1 for any given values of the Z;s.
it is sufficient to minimize:
subject to the above constraints. If any of the Z;s are zero, the
corresponding z;s must be identically zero. These Z;s are disregarded in what follows. Because of the constraints on the z;s, we
must introduce Lagrange multipliers and consider:
hfj= 9, E ?/iZPX2f(X)
+ 2: A($ z’
i=l
Y= 1 i
r=l
+ f:
i=l
1
.
p;~z'xf(x)
i
i=l
)
* = 1
A necessary condition for M2 to be at a constrained minimum
is that there exist a function X(X) (i.e., a separate multiplier for
each possible claim size) and C constants ,U; such that, for each
z; and each possible claim size:
L!!3=
4?f!JjZ”X2j(X)
+
2X(X)Zj
+
2/LjZjXf(X)
8Zi
= 2Zj[2$JjZfX2f(X) + X(X) + /LjXj(X)] = 0.
If z; # 0, then:
x(x> = -2z;x2f(,q- L'Xf(X)
@i
lcli
48
tLV.ANCING
‘lXANSA(“17ON
UXXS
AND
RISK I.OAI)
For a given claim size, we may sum over the insurers for which
Zj # 0 t0 get:
X(x)Xc G1 = -2x"/(x)-
&gl~xj(x).
k=l
ik #O
zk #()
Solving for X(x) and substituting back yields:
~az; = 2ZjXj(X)
If z; # 0, then:
X +
c
&2+k
k=l,Zk #o
7
z;x =
2”
I
k=l,zk#o
From the first necessary condition, we know that if Ml is at
a constrained minimum, an insurer that pays a given amount on
a claim of a given size must pay at least as much on claims of
all larger sizes. Thus, the number of insurers participating in the
payment of a claim may not decrease (and may very well increase) as the size of the claim increases. The difference function
of zfx with respect to x within a range of claim sizes with the
same participating insurers is:
zf(x + 1) - zfx =
c ll+i
c l/ljlk
k=l,Zk#o
BALANCING
TRANSACTION
COSTS AND
RISK LOAD
49
Since this expression is not dependent on x, we may conclude
that the optimal risk sharing arrangement must be constructed in
layers, with one insurer added at each successively higher layer.
Each participating insurer’s share of a particular layer is then
given by:
l/?cli
r
2
k=l,zk#O
l/?Ck
Recall that we have been assuming that Zi = E[zf. X] is fixed
for each insurer. Given the Zis, we now have enough information to determine in which order the insurers are added and the
boundaries between the layers without finding explicit values for
the pis. The highest layer has all of the insurers participating
with shares that have been determined above. The highest layer
boundary, ICPl, is determined by moving it down from K! until
the allocation of expected losses for one insurer, given by its
Zi, has been satisfied. That insurer is then dropped from further
participation and the next layer boundary down, 1~~2, is determined by moving it down from ICeI until the allocation of
expected losses for another insurer has been satisfied. This procedure is continued until all the layer boundaries have been determined.
The risk sharing arrangement described above minimizes A4
given that the expected losses allocated to each insurer are fixed
at certain amounts. It remains to find the risk sharing arrangement that minimizes M without any restrictions on the amount
of expected losses allocated to each insurer. To do this we must
find the optimal set of Zis. Each possible set of Zis is associated
with a set of layer boundaries, and vice versa. It is more convenient to focus on finding the optimal set of layer boundaries. The
optimal set of Zis will then directly follow.
50
BALAN(‘ING
TtIANSA(‘llON
4. OPTIMAL
(‘OSTS ANI1 KISK l.OA[)
LAYER BOINDARIES
Since we know that the optimal risk sharing arrangement
constructed in layers. we mav write M: as follo~vs:
is
where lo = 0, fc = 00, and G(.r) = 1 - F(?s) where F(x) is the
cumulative distribution function of X. At this point, we do not
know in which order the insurers should be added in successively
higher layers. The above expression, with insurers indexed according to the order in which they are added, could apply to any
ordering of insurers. Differentiating with respect to a particular
51
lj yields:
k=l
k=l
52
RALANCING
TRANSACTION
5$;/2+,)
-
r=l
COSTS AND RISK ILIAD
+ (v - 1). /:‘G(x)dx
+ I,
. 0
For Mi to be meaningful, we must have 0 5 f1 _<I1 < . <
Ice, 5 00. This will be referred to as the admissible region. The
first thing to note is that if any of the ljs are near infinity, MI
will not be at a minimum, since its derivative with respect to this
1; would be positive, thus indicating that ,441is increasing as 1;
approaches infinity.
For Ml to be at a minimum. the derivative of Ml with respect
to each f; in the interior of the admissible region must be zero.
We will now determine what conditions must be satisfied by any
Ijs at the boundary of the admissible region when M1 is at a
minimum. An fj is at the boundary of the admissible region if it
is coincident with another Ij or with zero.
First take the case where two or more 1jS are coincident with
one another at a nonzero point. 1,et the point of coincidence
be called & (s will be the index of the first of the ljs which
BALANCING
TRANSACTION
COSTS AND
53
RISK LOAD
coincide at Is). Suppose that y1of the ljs coincide at I,. Then, at
a minimum, the derivative of A41with respect to 1, must be zero.
If this were not so, all IZ of the ljs that coincide at 1, could be
either increased or decreased slightly to yield a smaller value of
Mi. The derivative of Mi with respect to I, is simply the sum of
the derivatives with respect to the n Ijs that coincide at 1,:
&$j,2$i)
%
= 2. t-q,).
a&
+ (v - 1). J” G(x)dx
0
i=l
+ I,
s+fl
-7(Qi/2$4+(v-1)
i=l
J0
s+n
i=l
C l/$i
= 2.G(f,).
i=s+l
i=s+l
s+n
c
l/$,i
i=s+l
+(v - 1). rlis G(x)dx
s
+ I, .
“G(x)dx+I,
54
I3ALANCING
‘I‘RANSA(“IlON
(‘OSI S ANI)
RISK II)A11
Note that the factor in brackets above is identical to the corresponding factor in the expression for the derivative of M, with
respect to a single lj except that $j+l is replaced by a weighted
average of $;+ls. If the II $j+ls corresponding to the n ljs are
not all equal to one another, then at least one of these @j+ls
must be smaller than the weighted average. If these insurers
are reordered so that an insurer with a $;+I smaller than the
weighted average is placed first, then if the derivative with respect to f, is zero (which implies that the factor in brackets must
be zero), the derivative of M1 with respect to the corresponding
li will be greater than zero. This implies that if this first lj of
those coincident at I, is moved down slightly. a smaller value of
A41 will result. Therefore, we conclude that, if M, is at a minimum, two or more 1;s may not be coincident with one another
at a nonzero point unless their corresponding o1 + Is are all equal
to one another.
We now move to the case where the first II of the 1;s coincide
at zero. If $,,. ..&,+l are not all equal to one another, then at
least one of these Qj+rS must be greater than or equal to all the
others, and strictly greater than at least one of the others. If these
insurers are reordered so that this insurer is placed last, then the
derivative with respect to the corresponding I, will be less than
zero. This implies that if this last li of those coincident at zero is
moved up slightly, a smaller value of MI will result. Therefore,
we conclude that if M1 is at a minimum, the first II 1,s may not
coincide at zero unless C$~= dz =
= d,,+ ,.
The above arguments imply that, if Mi is at a minimum, the
derivative of M, with respect to each I, must be zero. Therefore.
for each of the 1;s. the following equation must be satisfied:
I3AI.AN(‘IN(i
The relationship
lows:
‘Il<ANSA(‘l‘ION
(‘OS~I’S AND
IUSK l.OAD
55
between adjacent ljs may be expressed as fol-
It is clear from this equation that 1; - ii-1 will be positive
if and only if @j+, is greater than @j, and that lj will be equal
to fj-1 if and only if Qj+ 1 is equal to $j. Thus, to ensure a solution to these equations in the admissible region, the insurers
must be added in an order such that their @is are nondecreasing.
Furthermore, since the order in which insurers with identical $jS
are added does not affect the solution, there is only one solution,
which we conclude must yield the point at which MI, and hence
M, assumes its minimum value.
DISCUSSION
OF PAPER PUBLISHED
VOLUME LXXIV
IN
A NOTE ON THE GAP BETWEEN TARGET AND
EXPECTED UNDERWRITING
PROFIT MARGINS
EMIL10
DISCUSSION
C. VENEZIAN
BY WILLIAM
R. GIILAM
ACKNOWLEDGEMENT
Leigh Halliwell and John Rollins helped with the statistical analyscs. Leigh Halliwell derived the formula for the maximum likclihood estimate of 0’.
1.
INTRODUCTION
Dr. Venezian’s paper provides a simple yet powerful result:
the traditional actuarial pricing method produces an expected
underwriting profit margin that is lower than the target margin. This will not be avoided by an unbiased projection of losses;
as long as there is uncertainty in that projection. the results follow.
This uncertainty in the projection of loss costs is parameter
risk. The loading in rates for profir and conlirz~cncie.s should reflect the parameter risk assumed by the insurer, at least in the
contingencies part. Unfortunately. an appropriate loading for parameter risk is usually not susceptible to an easy statistical measure. Dr. Venezian’s theorem leads to a natural method for quantifying that loading. This review uses that method to calculate a
contingencies loading for workers compensation rates.
The reviewer is aware of the controversy surrounding Dr.
Venezian’s result, and has read Sholom Feldblum’s review [l]
56
TARGET
AND
EXPECTED
UNDERWRITING
PROFIT
MARGINS
57
several times. Mr. Feldblum does not refute the statistical theorem
1
1
(1)
E x ’ E[X]
[1
for positive-valued random variable X, but points out that the
correct way to combine the loss ratios of several individual policies is not to take a straight average but to aggregate them; i.e.,
add all the losses for the numerator and all the premiums for
the denominator. This is equivalent to a weighted average of loss
ratios. If, as in Feldblum’s example, the many policy loss ratios
encompass the complete distribution of projection errors, there
is no projection error left. The variance in loss ratios by policy is
irrelevant.
This reviewer would not be so quick to dismiss Venezian’s result. The theorem can still be applied to loss ratios that cannot be
aggregated. For sake of this discussion, the reviewer has had to
decide when a loss ratio must stand on its own. Individual company calendar quarter loss ratios, for instance, do not hold much
actuarial relevance. Nevertheless, they seem to generate a fair
amount of discussion in financial markets and among carriers.
The reviewer has selected a one policy year statewide loss ratio
as having enough actuarial and financial relevance to stand on
its own. The bureau estimates and files a (pure) premium level
change by state each year. In order to realize a certain underwriting return on premium, admitting that the filed loss rates are
an estimate, we wish to determine how they should be loaded for
profit and contingencies.
Alternatively, the analysis could be done by grouping states or
years, which would result in a smaller, but non-zero, load. Larger
companies can combine a few states before calculating results,
but there are many companies writing in one or two states that
cannot afford this luxury: Parameter risk affects the bottom line
results.
2. DEVELOPMENT
OF ALGEBRA FOR ANALYSIS
RATE LEVEL CXCERTAINTY
OF
As each renewal date approaches, the actuaries must choose a
single estimate of the needed rates in the ensuing year. In workers compensation, the indicated rate change is estimated as a
projected ratio of loss to premium at current level. divided by the
permissible loss ratio. or PLR. Once a rate change is approved,
the actuaries can revise their projected loss ratio to reflect the
actual rate change. When the year is complete, and the actual
premium is reported, projected losses, or PRJ. are calculated by
extending that premium by the revised projected loss ratio.
The emerged actual losses for the year are a random variable ACT. with some unknown expected value TAR so that
E[ACT] = TAR. The quantity name TAR is used to evoke
Stephen Philbrick’s article on credibility [2]. Philbrick likened
the estimation of an unknown parameter such as T,4R to target shooting. The value of ACT varies around TAR because of
the random nature of the process, the process variance. At any
time before maturity, the exact value of ACT is unknown and
must be estimated by actuaries.
PRJ is also a random variable, the outcome of a stochastic
process called ratemaking, based on data subject to errors, formulas subject to assumptions. and debate prejudiced by politics.
For purposes of this exposition, we assume the loss projection is
unbiased, thus E[PRJ] = E[ACT] = TAR.
We define a random variable X by:
X=PRJ
TAR’
so
PRJ = TAR* X.
(2)
X is a positive-valued random variable, with non-zero variance.
By its definition, E[X] = 1.
TARGET
AND
EXPECTED
IJNDliRWRIT~N(i
PROF’IIT MARGINS
59
As stated by Dr. Venezian,
1
E[x11>EIXl=l
for any positive valued random variable X with unity mean. Except in some degenerate cases, strict inequality will obtain. (The
variable X here is the same as Venezian’s 1 + X.)
Following Venezian’s logic, to assure realization of the profit
provision, rates should be multiplied by the factor E [+I; or, alternatively, the PLR (in the original filing) should be divided by
E [ $1. This adjustment should be made after the target profit
provision is established using cost of capital and/or other economic evidence.
In practice, the loading would be an element added to the
provision for expenses, tax and profit (the complement of the
PLR). To develop that loading. define a new target loss ratio,
PLR’. Then:
=PLR-PLR(E[;]-1).
The element added to expenses is then PLR (E [+] - 1). It will
be largest when the uncertainty in the projected loss costs is
risk is greatest.
greatest; that is, when the pururneter
3. ESTIMATION
OF THE COKTINGENCY
LOADING
To estimate parameters of the distribution of X for workers
compensation statewide rate level indications, the reviewer has
60
TARGETANDEXPECTED
~NDER~KITINGPKOI;I.~MAKGINS
assembled reported financial data comprising eight policy years’
loss ratios for twenty-six states. These loss ratios are developed
to ultimate as of the latest evaluation at 12/31/92.
For each state and policy year, there is also a projected loss
ratio based on an analysis of rate indications and approvals as
described in Section 2. Weighted averages must be taken in cases
where rate changes occur at other than January 1.
The general approach is to compare the actual emerged losses
by policy year with those projected at the time of rate level
approval. The projected losses are the product of a projected
loss ratio and earned standard premium. The quotient, projected
losses divided by actual losses, will be used as a sample estimate of the random variable X. This requires several assumptions documented below. From the many samples, statistics of
the distribution of X are derived.
Exhibit 1 displays two of twenty-six states’ data used in this
estimation. Calculations progress from left to right, across the
page in the usual fashion. The reader should anticipate the eventuality of looking (down) through the pages (through the states)
to calculate statistics pertaining to all states in each policy year.
The notes below explain each column.
1) Policy years 1984 through 1991 are used.
2) Standard premium shown is as actually earned.
3) Projected Loss Ratio is that actually expected given the rate
filing approval.
4) The Projected Loss, PRJ, is a product of Actual Standard
Premium (2) and the loss ratio expected after the rate change
(3).
5) Incurred Loss is as of the latest evaluation, developed to
ultimate. This is a best estimate of ACT. We will be using ACT
as an estimate of TAR.
TARGET
AND
EXPECTED
IJNDERWRITING
PRDFTT MARGINS
61
6) The ratio (4) t (5) is the ratio of the projected to actual
losses (which is also a ratio of loss ratios to on-level premium).
PRJ/ACT is an estimate of the random variable PRJ/TAR = X
defined above. The denominator, ACT, is an estimate of the underlying targeted losses, TAR. This estimate is subject to two
principal errors-process variance and error in the estimated development to ultimate. The process variance we may safely disregard as small using the following logic:
The emerged losses ACT in Column 5 vary around some true
expected value TAR, (by year y) due to process variance. Ignoring estimation error for a moment, variance of PRJ/ACT will
be greater than the variance of PRJ/TAR, but by an insignificant
amount. We can estimate the variance of L = ACT/TAR using
risk modeling concepts. It has a relatively small variance.
Assuming frequency and severity are independent,
Var[L] = Var ( EPACCTT])= ELv1va~$~~~1E]z12,
(3)
where y is the claim count and z is the severity random variable.
so:
L[36
+ 21,
z VYI
using reasonably conservative estimates of the variance components. (If these were doubled, it would not change the conclusion
that process variance is relatively small.) Then
Var[L] = &
= .0004,
,
where 100,000 is clearly a low estimate of expected claim count
in almost any state.
The error in development to ultimate is probably more significant, but is at least of the same nature as error in the original
62
TARGET
AND
liXPl:(“l‘iiD
IJNI)I:1<WI~I’fIN(i
I’HO1~l’l’ MARGINS
projection. The basic quid pro quo for being unable to unravel
this estimation error is that so much of the variance of the projection is eliminated in the next step.
7) Because this contingency loading is not a correction for
bias in the projection of losses, the estimates in Column 6 have
been normalized by state so that over the eight years in the sample, the average error of the projections is nil; i.e., the ratios
average to unity. This effectively ignores a lot of parameter risk
exhibited in the data, probably of a much greater magnitude than
whatever parameter risk is introduced by the immaturity of the
evaluations. Even after this adjustment, a significant amount of
error remains, and we will try to estimate its distribution. Elements of Column (7) are sample estimates x; of the variable X.
The average over the eight years. which is now unity. is shown in
the last row.
8) We have observed above that the loss projection process
is a stochastic process, the result of multiple judgments. Most of
these judgments are factors-factors
for loss development. trend,
law evaluations, etc. It is natural to use a lognormal distribution
to model the results of such a process. With the goal of fitting a
lognormal, we take logarithms t, = Inx; of the sample points xi
in Column 7 and square them. The l!z will be used to estimate
the parameter c2 of the lognormal distribution.
The ninth row shows the averages of each of the Columns 6,
7, and 8.
Since we are estimating the parameters of a distribution with
mean of unity. we can require that p = -;a’, so that the lognorma1 mean will be unity. We must then est;mate only the parameter g2 from the sample. The maximum likelihood estimate S’ of
the parameter $ is given by the following:
(4)
This leads to by-state evaluations of S” in the tenth row. For
each state, a contingency loading, (e”? - 1) * PLR is shown in the
eleventh row. A better estimate of gz is the calculation of S’
across all the states for each year. This is calculated in the next to
last column of Exhibit 2 using the respective elements of Column
7 from each state. The values of S2 vary from 0.0036 to 0.0548
over the eight years in the study.
The statistic S2 = 0.0214 calculated in the first half of Exhibit 2 uses all 208 (= 26 x 8) estimates of X in the exhibit.
For the lognormal distribution of X with parameters [-in2,g2],
E [=$I = e6’. When g2 = 0.0214, E [$] = e0,“214g 1.022. This
leads to a contingency loading of PLR(1.022 - 1) G’ 1.5%, when
the permissible loss ratio is 70%.
For the record, the second half of Exhibit 2 calculates a contingency loading when there has been no normalization of projection error. This is about 4%.
When the risk can be spread over more states the loading
could be lower. It has also been suggested that the loss ratio
should be aggregated over more years, and the loading thus reduced. As long as ratemaking is an inexact science, the loading
should be non-zero.
64
‘IxI<(iI!t’
AND
t~Xt’IiC’l‘t3Zl
IJNtlI~tIWI<t’I‘1N(i
1’11OI~t’t’ MAKCiINS
REFERENCES
[l]
Feldblum, Shalom, Discussion of Venezian: “A Note on the
Gap Between Target and Expected Underwriting Profit Margin,” PCAS LXXVII. 1990, pp. 42-95.
[2] Philbrick, Stephen W., “An Examination of Credibility
cepts,” PCAS LXVITI, 1981, pp. 195-219.
Con-
EXHIBIT
1
PART 1
CALCULATION OF CONTINGENCY LOADING
STATE A
(1)
(3)
(2)
Policy
Year
I
Standard
Projected
Premium
Loss Ratio
200,278,065
62.3%
61.9%
256,463,153
316,065,139
62.7%
358,210,729
63.8%
64.4%
389,240,500
453,685,090
64.6%
437,795,706
72.9%
74.2%
1991 1 420,210,734
(4)
(2)*(3)
Projected
Losses
124,773,234
158,750,692
198,200,656
228,450,683
250,561,895
293,272,502
3 18,953,543
311,814,224
T
(3
(6)
(7)
(8)
(4)/W
Wavg(6)
(WY
Initial X
Balanced X MLE
Incurred
Estimate
Estimate Summand
Losses
0.7115
0.9216
0.0067
175,356,228
0.6793
0.8798
0.0164
233,709,184
0.7554
0.9784
0.0005
262,386,482
0.8006
1.0369
0.0013
285,366,347
354,792,5 10
0.7062
0.9147
0.0079
38 1,609,365
0.7685
0.9954
0.0000
0.7966
1.0317
0.0010
400,409,757
325,3 16,985
0.9585
1.2415
0.0468
Maximum Likelihood Estimator (S) = L
PLR<
0.706h
0.0100 c
Indicated Contingency Loading = pi?iq
(exp(Sz)-l)*PLR
EXHIBIT
1
PART 2
CALCULATION OF CONTINGENCY LOADING
STATE B
(1)
Policy
Year
(2)
(3)
Standard
Projected
Premium
Loss Ratio
140,918,339
63.0%
141,994,308
63.5%
136,240,676
62.9%
157,438,852
63.8%
192,055,479
62.7%
224,943,549
62.7%
217,048,436
64.4%
228,893,562
64.5%
bnwtd AVB
(4)
m*(3)
Projected
Losses
88,827,393
90,166,386
85,682,919
100,415,130
120,418,785
141,039,605
139,779,193
147,636,347
(7)
(8)
(6)
(6)/avg(6) (ln(7))’
(4)/(5)
Initial X
Balanced X MLE
Estimate Summand
Estimate
- --~0.8661
0.7438
0.0876
0.8969
0.7702
0.0682
0.9889
0.8493
0.0267
1.1977
1.0286
0.0008
1.5422
1.3244
0.0789
1.4101
1.2110
0.0366
1.2311
1.0573
0.0031
1.1824
1.0154
0.0002
(5)
Incurred
Losses
102,565,294
100,536,644
86,646,7X7
83,836,649
78,084,105
100,022,404
113,539,367
124,865,150
I
1.1644
1.oooo
0.0378 b
Maximum Likelihood Estimator (9) = ~---o.o3741(
Indicated Contingency Loading = /y-mq
(exp(S*)-l)*PLR
EXHIBIT
1
PART 3
CALCULATIONOF CONTINGENCYLOADING
STATEC
(1)
(2)
(3)
(5)
(6)
Projected
Losses
Incurred
Losses
(4)/(5)
Initial X
Estimate
163,150,250
230,659,459
285,876,133
313,409,179
337,409,506
375,471,061
411,530,956
2 19,855,325
234,7 16,774
270,424,555
308,069,073
357,625,450
374,780,756
413,304,435
(4)
(2,*(3)
I‘olicy
Year
1987
1988
1989
1990
Standard
Premium
26 1,040,400
369,646,569
462,582,740
5 15,074,069
566,185,080
606,576,835
657,397,693
Projected
Loss Ratio
62.5%
62.4%
61.8%
60.8%
59.6%
61.9%
62.6%
0.742 1
0.9827
1.0571
1.0173
0.9435
1.0018
0.9957
(7)
(8)
(6Yavg(6) (ln(7))’
Balanced X MLE
Estimate Summand
0.7623
1.0095
1.0859
1.0450
0.9692
1.0291
1.0228
0.0737
0.000 1
0.0068
0.0019
0.0010
0.0008
0.0005
Maximum Likelihood Estimator (S’) = pGiiiq
Indicated Contingency Loading =
(exp(S?)-l)*PLR
EXHIBIT
2
PART 1
CALCULATIONOF CONTINGENCY LOADING
X ESTIMATES NORMALIZED OVER POLICY YEARS
STATE:
A
-ii
0.9216
0.7438
__
-0.8798
0.7702
.~_.
0.9784
0.8493
1.0369
1.0286
__
1988
0.9147
!:3244
1989
0.9954
1.2110
__.~__~~~___
1990
1.0317
1.0573
1991
1.2415
1.0154
PY:
1984
1985
1986
-1987
I Tvc-P-r.oooo
1.0000
C
D
-..l.~
0.7623 .._..__
0.8869
----.
1.0095
0.95 13
0.8172
!.ossS
1.0450
0.8849
E
1.1806
-..0.98 15
1.0833
0.8982
P
0.7664
0.8680
0.8730
0.8634
c
1.4200
1.2845
1.2605
0.9954
H
0.8322
0.9056
0.9424
0.9954
I
0.9226
0.9161
0.9955
1.0571
J
0.9722
1.0247
1.0087
0.9676
K
0.8613
0.9098
1.0210
0.9597
L
1.1411
1.1479
1.1354
1.0579
M
0.6027
0.5983
I.0118
1.0574
N
1.0160
0.9881
0.8783
1.0181
0.9692
! ,029 1
I .0228
1.0762
0.9625
0.9730
1.1688
1.3553
0.93 13
1.0480
0.9560
0.9211
0.7993
1.0232
1.4859
1.32 IO
0.8353
0.7771
0.7071
0.7202
1.0103
0.9817
1.0957
1.2365
1.0232
0.9701
1.0436
1.0718
0.9505
0.9216
1.1045
1.0503
1.0292
0.9735
I .0576
1.1879
0.9920
0.8 I23
0.8292
0.8843
1.1030
0.8974
1.2197
15097
1.0210~
0.9185
1.0141
1.1460
l.oow
1.oooo
1.0000
1.m
1.oooo
I.oooo
I.0000
1.0000
1.0000
l.ooOO
I.ooW
l.OOoO~
EXHIBIT
2
PART 2
CALCULATIONOF CONTINGENCY LOADING
X E~TIMATESNORMALIZED OVERPOLICYYEARS
PY:
1984
1985
1986
1987
1988
1989
1990
1991
AVC
STAT
LOAD
0
0.9809
1.0587
1.1376
I.0165
0.9948
0.9289
O.PlC-4
0.9722
P
0.5008
0.6958
0.8750
0.9860
I.1099
1.2173
I.2317
1.3837
P
I.1290
1.0928
0.9567
0.9544
0.8732
0.9347
0.9868
1.0724
R
0.8613
0.9824
1.0829
1.0698
0.9456
1.0622
0.9622
1.0336
s
0.7806
0.8379
0.9834
0.8961
0.9773
1.1424
1.0440
1.3384
T
1.0722
1.0204
1.0978
I.0076
0.9932
0.8449
0.9301
1.0336
II
0.9866
I.0191
1.0655
0.9961
0.9468
0.9907
0.9480
1.0472
V
0.8531
0.9484
1.1633
1.0409
1.1639
1.0233
0.9738
0.8333
W
0.8014
0.8930
0.9229
0.9292
I.0071
1.1748
1.0940
1.1777
--X
1.0465
1.0676
1.0762
0.9866
0.9506
0.9290
I.0094
0.9342
Y
I.1154
10567
0.9499
1.0377
0.9715
1.0408
0.9830
0.8451
z
0.8316
0.9315
0.9178
l.OlOa
1.0465
I.0599
1.0677
I.0751
1.0000
0.0044
031%
l.OOoO
0.1008
7.42%
IO000
0.0068
0.48%
1.0000
0.0053
038%
1.0000
0.0261
Ias%
l.OGaO
0.0061
0.43%
m-JOO
0.0016
0.11%
l.OOoO
0.0137
0.97%
l.OOaa
0.0167
l.lwm
I.OOOCl
0.0031
0.22%
I.OoOO
0.0062
0.44%
l.OOM
0.0068
0.48%
--AVC
0.9227
0.9554
1.0088
0.9922
0.9941
0.9954
1.0360
STAT
0.0548
0.0261
0.0109
0.0036
0.0098
0.0119
0.0185
LOAD
3.95%
1.85%
4.77%
0.25%
0.49%
064%
131%
EXHIBIT
2
PART 3
CALCULATIONOFCONTINGENCYLOADING
X ESTIMATES NOT NORMALIZED
_.
1985
1986
I987
1988
1989
199@
199lJ
I
0.9827
I.0571
1.0173
3.9435
1.0018
0.9957
1.0477
0.7042
0.6049
0.6550
0.7 124
0.7202
0.8652
1.0032
I m93
I.1141
0.9237
0.x77
1.0777
0.9832
0.9472
0.7 146
0.7187
0.7108
0.6580
0.8423
1.2233
I.0875
1.5270
I.4986
1.1834
0.9931
0.9239
0.8406
0.8562
0.8563
0.8352
0.7554
o.w3ob
0.7062
0.7685
0.7966
0.9585
0.8969
0.9989
1.1977
1.5422
I.4101
1.2311
1.1824
0.8911
o.V412
0.9553
0.9283
1.0361
I.1692
09076
0.%38
0.9328
0.8844
0.95 15
0.9772
0.833b
01721
0.078 I
5.69%
l.lb44
0.0545
3.92%
0.9735 .__.._
0.7402
0.0122
0.1184
0.86%
8.00%
1.0284
0.0083
03%
0.8233
0.0980
7.21%
1.1889
0.0861
6.32%
0.9456
O.Olb4
1.16%
0.9117
0.0120
0.84%
0.6793
0.8189
0.7V40
0.9515
0.9048
0.7745
0.8692
0.8170
0.8761
0.8288
0.9004
I.0113
0.9675
0.9569
0.8916
0.8360
0.6846
0.6989
0.7453
0.4862
0.8221
0.8592
0.8962
0.7292
0.9911
1.2267
0.83bi
0.7438
O.Rb22
0.8646
0.7778
0.8588
0.9705
0.8615
0.0255
1.81%
0.8513
0.0354
2.52%
0.8428
o.oso4
3.62%
0.8126
0.1500
11.33%
0.8469
0.0338
2.40%
0.8828
0.8690
EXIIIBIT
2
PART 4
CALCULATIONOFCONTINGENCY LOADING
X ESTIMATES No'r NORMALIZED
STATE:
0.7844
0.7997
05219
09306
0.7761
0.6743
0.8179
07382
1.2687
07704
0.9570
0.9868
0.6445
0.7412
1.1562
0.7884
I.0423
0.8242
0.9010
0.8639
0.7628
0.9271
06185
fl.8300 - 0.9338
0.9228
1.2649
0.7773
07540
I.1613
0.7835
0.9056
0.9089
07410
0.7482
10868
0.7718
07931
0.8132
0.8371
AVC
08268
STAT
0 1048
DISCUSSION
OF PAPER PUBLISHED
UNBIASED
IN VOLUME
LOSS DEVELOPMENT
DANIEL
DISCUSSION
LXXX1
FACTORS
M. MURPHY
BY DANIEL
E GOGOL,
P1i.D.
Daniel Murphy presents some powerful and useful techniques
for estimating biases and variances of loss reserve estimators. It
is mentioned in the introduction of Murphy’s paper that Casualty Actuarial Society literature is inconclusive regarding whether
certain loss development techniques are biased or unbiased. It is
also stated that the paper provides a model so that these questions, and others, can be answered.
Although the assumptions of Murphy’s models enable him to
show that the simple average development factor method and the
weighted average development factor method are unbiased, actually they both are biased upwards. (The Bornhuetter-Ferguson
and Stanard-Biihlmann
methods also are biased upwards if they
use these factors.) It is only because Murphy’s models have unrealistic properties that it is possible to prove that the estimators
are unbiased. Murphy is aware of this, as is shown by his discussion of claim count development in Appendix B. He states:
Take the weighted average development method for example. Clearly there is a positive probability (albeit
small) that x = 0, so the expected value of the weighted average development link ratio J~/X is infinity.
Murphy also indicates that a general, heuristic argument that
weighted average development yields biased estimates can be
found in Stanard [3]. Stanard’s argument for the bias of weighted
average development factors [3, Appendix A], is actually only a
derivation of an equation which must be satisfied in order for
the factors to be unbiased. Stanard states without proof that the
equation is not true in general. (It can be seen that the equation
72
UNBIASED
LOSS DEVELOPMENT
73
FACTORS
is untrue by considering Murphy’s above point; i.e., the positive
probability that x = 0.)
Murphy’s statement that the expected value of the weighted
average development link ratio is infinity is true; but of course,
further analysis is necessary to give some idea of the amount of
the bias.
In actual practice, a reserving actuary would obviously not use
an infinite link ratio. In many situations the possibility of an infinite weighted average link ratio is remote and it may be judged
that any weighted average link ratio greater than some R will be
replaced by R in computing weighted average development factors. Suppose also that if the weighted average link ratio is O/O,
the zeroes will be replaced judgmentally by some y/x.
However, even with this new system, weighted average development factors will be biased upwards if the probability that
y/x is O/O or y/x > R is sufficiently small. The proof that will be
given could be useful in estimating the amount of the bias.
The case of a single factor for a single accident year will
be considered first although the same argument applies to policy years or report years. It will later be shown that the result demonstrated for a single factor applies to weighted or unweighted averages of factors.
Let X and Y be random variables which represent the reported losses for an accident year at evaluations x and y years
after the start of the accident year. The factor which, when multiplied by X, produces an unbiased estimate of the mean of Y is
E(Y)/E(X),
since
W(WYWXNX)
= QW’)/E(OWX)
= JV’).
It is not true in a realistic model that, given a particular value
x necessarily equals, or is even a good
x of X, (E(Y)/E(X))
approximation of, E(Y 1X = x). For example, it is not true that
E(Y 1X = 0) = (E(Y)/E(X))(O)
= 0.
74
UNBIAStD
I.OSS DbVl
I.OP.11 Y 1 1,.4(‘ lOI<S
The equality (E(Y)/E(X)).l= E(Y 1X = s) is. however, implicitly assumed in Murphy’s Model Il. which he described as
follows: )’ = hx + e. E(e) = 0, Var(c) is constant across accident
years, and the e’s are uncorrelated between accident years and
are independent of .r.
In a realistic model, a prior expectation E, could be estimated for Y, based on data other than the loss experience of
the accident year being considered. E(Y ] X = my)can be approximated by a credibility weighting of E, and the experience indication (E(Y)/E(X))x.
(See [3].) The demonstration that will
be given of the upward bias in development factors uses the assumption that, for x > 0. E(Y 1X = .r) is equal to a weighting
of the form (1 - Z)E, + (Z)(E(Y)/E(X))n-.
However, it can be
seen that all the steps of the argument also hold true if the above
weighting is a good approximation as is generally true in actual
practice.
It follows from the above assumption that
E(Y 1X = x) > (E(Y)/E(X)).y
if x is sufficiently less than E(X). Similarly.
E(Y ] X = x) < (E(Y)/E(X)).r
if x is sufficiently greater than E(X). Also, it is clear that
E(Y 1X = x)/ x is a monotonically decreasing function of \-.
Let X’ and Y’ be the random variables \vhich represent the
values of X and Y. respectively, after they hav,e been judgmentally changed, as described previously, if X = 0. If the probability
that X = 0 is sufficiently small, then E(Y’)/E(X’)
is ver!’ close
to E(Y)/E(X).
It will be shown that
E(Y’ ] X’) > E(Y’)/E(X’).
(‘1
Let S(x) be the probability density function of X’. Then
J
‘=(E(Y’/X’)
0
] X’ = x)xf(x)dx
= E(Y’).
tJNt3tASI:tl
I.OSS tXVIII.Ol’MIW’t’
7s
I~ACTOIIS
Therefore,
J
33(E(Y’/X’)
0
I X’ = x)(x/E(X’))/(x)dx
= E(Y’)/E(X’).
(2)
However,
E(Y’/X’)
= i=(E(Y’/X’)
I X’ = x)/(x)dx
>J
I X’ = x)(x/E(X’))f(x)dx,
=(E(Y’/X’)
(3)
0
as will be shown below. It follows from (2) and (3) that (1) is
true.
In the two integrals in the above inequality, the same function of x, i.e., E(Y’/X’) I X’ = x, is multiplied by S(x) and by
(x/E(X’))f(x),
respectively.
It was mentioned above that f(x) is the probability density
function of X’. The function (x/E(X’))f(x)
is also a probability
density function, since
~x(x/E(X’))j(r)di
= (l/E(X’))i-
xS(x)dx
= 1.
Note that:
WV’NSW
< J(x)
for
x < E(X’),
wvo)m)
> P(x)
for
x > E(X’).
and
Thus, the density function (x/E(X’))/(x)
gives less weight than
f(x) to values of E(Y’/X’) ) X’ = x for which x < E(X’), and
more weight to values of E(Y’/X’) I X’ = x for which x >
E(X’). However, E(Y/X) 1X = x is a monotonically decreasing
function of x. If X’ and Y’ are not too different from X and Y,
then E(Y’/X’)
I X = x is close enough to a monotonically decreasing function so that Equation 3 is true. This completes the
proof of upward bias for a single factor.
76
IJNBIASED
LOSS DEVELOPMENT
I’AC’I‘OKS
Murphy uses the assumptions that the expected development
pattern for each accident year is identical, and that development is independent for each accident year. It follows that the
expected value of a simple average of development factors, at
the same evaluations, equals the expected value of any individual factor. Therefore, using Murphy’s assumptions and the proof
above, the simple average of development factors is biased upwards.
It also follows from Murphy’s assumptions that weighted average development factors are biased upwards. For a set of accident years, let X; and Yi represent the reported losses for accident year i at evaluations x and y years after the start of accident year i. Then the random variable Y/X, where Y = CYi
and X = C Xi, equals the weighted average development factor.
It can easily be verified that the proof of Equation 1 is valid with
the previous definitions of X and Y (and the corresponding X’
and Y’) replaced by these new definitions.
UNBIASED
LOSS DEVELOPMENT
FA\CTORS
77
REFERENCES
[l] Gogol, Daniel, “Using Expected Loss Ratios in Reserving,”
Insurance: Mathematics and Economics 12, 1993, pp. 297-299.
[2] Murphy, Daniel, “Unbiased Loss Development
PCAS LXXXI, 1994, pp. 154-222.
Factors,”
[3] Stanard, James N., “A Simulation Test of Prediction Errors
of Loss Reserve Estimation Techniques,” PCAS LXXII, 1985,
pp. 124-148.
DISCUSSION OF PAPER PUBLISHED
VOLUME LXXVII
IN
RISK LOADS FOR INSURERS
SHOLOM
DISCUSSION
FELDBLUM
BY STEPHEN
VOLUME
AUTHOR’S
LXXVIII
REPLY TO DISCUSSION
VOLUME
DISCUSSION
1.
PHILBRICK
LXXX
BY TODD BAULT’
INTRODUCTION
I have been following with great interest this discussion
“thread” in the Proceedings [I, 2. 31, along with the recent papers of Rodney Kreps [4] (with Daniel Gogol’s reply [5]) and
Glenn Meyers [6] (with Ira Robbin’s reply [7] and Meyers’s response [S]). Obviously, this is an important topic for the CAS,
as evidenced by the amount of discussion it has generated; and
it is of particular interest to me, given my current specialization
in rate of return, surplus need, and related areas of financial actuarial practice. The focus of the FeldblumiPhilbrick discussion
has been five methods of setting risk loads and the relative merits and deficiencies of each. The other papers by Kreps and by
Meyers deal with related approaches and issues. I wish to add
two observations to the discussion:
‘The author would like to thank Mr. Randall Hoimbcrg for the training, insight, and encouragement hc has provided mc over the years , and for the many stimulating discussions
we have had. one of which led to this paper.
78
l
l
All of the methods are more similar than different, including
the methods discussed by Kreps and Meyers, and if care is
taken to use a common set of assumptions, the methods are
nearly equivalent;
None of these methods. including
CAPM, resolves several fundamental problems; and any risk loads derived from these
methods must still contain a great deal of subjectivity, more
than is implied by Feldblum’s discussion of CAPM.
I wrote this discussion to question the level of certainty conveyed in Feldblum’s initial paper and to keep the topic open. AIthough Philbrick’s comments help in this regard, he does not go
far enough. I am concerned that inexperienced actuaries will see
betas published in the Proceedings and will feel justified in rushing off to use these in setting profit loads, despite Feldblum’s
warning that his calculations were for illustration only. There are
still many unresolved issues regarding the measurement of risk
and its application to profit provisions. The research to date is
encouraging and highly connected, as we shall see, but there is
still much left to do.
2. THE FIVE RISK LOADSARE
THEY
EQUIVALENT?
I start this analysis with the work of Rodney Kreps-his
paper already describes most of the connections I want to demonstrate, but they have not been fully integrated. We can use his
equations to show that variance, standard deviation, ruin theory,
and CAPM describe similar (and nearly equivalent) concepts.
Although I could not incorporate utility theory and the reinsurante method with sufficient mathematical rigor, there is reasonable evidence that these latter approaches are also strongly related to the others.
We begin with Kreps’s equation for surplus supporting insurance variability of a given portfolio. Kreps assumes this portfolio
represents a company’s book of business, and Feldblum agrees
80
RISK I.OADS
FOR INSIJRERS
that this assumption is appropriate for adapting CAPM to insurance. However, I will assume this portfolio is the industry portfolio (in the next section, I will support this position-ire
could
use a company portfolio in this analysis and reach the same conclusions). The equation is
V=zS-RR,
(2.1)
where V is the surplus, S is the standard deviation of the portfolio. R is the return in dollars, and z is the standard normal
percentile value associated with a given probability of ruin (i.e.,
exceeding needed surplus). Kreps does not show explicitly that a
ruin theory equation produces this formula; it is
Pr(L + E > E(L) + E + R + V) < e,
(2.2)
where L is the random variable for loss (boldface will always be
used for random variables), E is expense, and e is the threshold
probability of ruin corresponding to the value z. Note that the
standard deviation of L is S. Standardizing I, to (L - E(L))/S
produces the solution for z, from which (2.1) follows:
z = (R + V)/S.
(2.3)
Kreps then produces the equation for the marginal surplus required for a new risk X. We shall assume x is very small in magnitude compared to L, for both means and standard deviations.
The equation for the marginal surplus is
V’ - V = z(S’ - S) - r,
(2.4)
where V’ and S’ are the surplus and standard deviation, respectively, for the portfolio with x added, and r is the return for X.
Further, Kreps solves for S’ - S as
9-s
= a(2SC +cT)/(s'+
s>.
(2.5)
where g is the standard deviation of x, and C is the correlation
coefficient of x and L.
Gogol, in his discussion of Kreps’s paper, noted that this approach is highly “order dependent.” That is, Equation 2.5 shows
RISK LOADS
FOR INSURERS
81
the increase in standard deviation if x is the last risk added to
the portfolio. If we assume that L is a portfolio of risks identical
to X, then x’s contribution would be c if it were the first risk in
the portfolio, a smaller number than g (or at least not larger)
for the second risk, and so on, despite the fact that the risks
are identical. Measuring each risk’s surplus requirement based
only upon its marginal risk contribution will underestimate the
total surplus need of the portfolio. Gogol developed a formula
to allocate the total surplus need to all individual risks based
upon an average of the risk’s standard deviation on a “first-in”
and “last-in” basis. (Please see Gogol’s paper for more details.)
This is an important adjustment for practical implementations of
this method, but please note that Gogol’s technique is not the
only way to do this-another
approach will be discussed at the
end of this section.
Up to this point, we could call Kreps’s method a ruin theory
approach, because ruin theory is the basis of his equations. But
is this method related to any other approaches? Suppose that x
and L are independent (rare, but the usual assumption), so that
C = 0. Then
s’ - s = a2/(S’ + S) E a2/(2Q
P-6)
because Q is small compared to S. Thus, the marginal surplus is
a function of the variance of the new risk. (Feldblum cites this
formula in Footnote 1 of his reply to Philbrick, but does not
mention explicitly the independence requirement.) Conversely, if
x and L are completely dependent (also rare, but illustrative),
then C = 1 and
s’ - s = a(2S + a)/(9
+ S) iz CT.
(2.7)
Again, this is true because 0 is small compared to S. Now the
marginal surplus is a function of the standard deviation of the
new risk. So in the most common situation, where x is slightly
correlated with L, the marginal surplus will be a linear combination of the variance and standard deviation related to the covariante. In my opinion, this makes the whole “variance vs. standard
deviation” debate much less interesting, because both are simply special cases of a unifying covariance framework. Actuaries
may continue to choose one or the other method on the basis
of tractability concerns (and measuring covariance is very difficult), but they should be aware of what these decisions imply and
whether or not their assumptions are appropriate.
We have two important results so far:
1. The distinction between variance and standard deviation
methods is somewhat artificial. Which method to use is
a function of the correlation between the new and existing risks, and in most cases. the “correct” answer is a
marginal risk approach that incorporates covariance.
2. Marginal risk methods (including variance and standard
deviation methods as special cases) are closely related to
a ruin theory approach.
Let us examine S’ - S further. Define P as the premium associated with the industry portfolio, and let p be the premium
associated with the new risk. From Equation 2.2, it should be
clear that S/P is the standard deviation for the industry return
on premium. Further, (S’ - S)/p is the marginal contribution
to
the standard deviation of the return on premium from the new
risk. Using Equation 2.5 and some algebraic manipulation, note
that
KS’- wPIIwpl
= [(2SCa
= [@SC + M(S’ + s>x P)lI[~IPl
+2)/(p
x 2P)]/[(S/P)x
= [C x @lP>x www/p)
(because d/(pP)
(s'+s)/2p]
x (S/VI
is small)
= cov(x/p,L/P)/var(L/P).
WV
The last part of Equation 2.8 looks remarkably like a CAPM
beta. In fact, this is the formula for beta proposed by Feldblum,
so let us “set” /3 equal to [(S’ - S)/p]/[S/P].
This is not how
RISK LOADS
FOR INSIJRERS
83
beta would be derived in practice, but it serves as a link in the
chain of reasoning of this analysis. However, Feldblum might disagree with this characterization, as I have just said, in effect, that
the variance in profit equals the variance in loss, and Feldblum
produced at least three examples to demonstrate that this is not
true. Before proceeding, then, I should justify this simplification.
Two of Feldblum’s examples are rather naive descriptions of
how to measure the variance of losses, and it is fairly easy to
remedy the problems he describes. His first example, retrospective rating, could be fixed by measuring the variance of the insurer’s effective loss distribution, which is zero in his idealized example. His third example, heterogeneous mix of risks, is fixed by
using homogeneous groups, or by adjusting for the heterogeneity
in a reasonable fashion.
But Feldblum does indicate some important sources of risk
that are not derived from the loss distribution, including inflation, investment, default, and parameter risk. All of these need
to be measured, but the method for doing so does nut require
measuring the variance of profits directly. In fact, given the problems with calendar year measures of profitability in insurance
(which Feldblum used, although he did acknowledge that problems existed), it may be preferable to measure the variance of
profitability in other ways, such as starting with the variance of
accident year losses and modeling additional sources of risk as
required. (This will be discussed further in the next section.)
Thus, as long as x and L reflect these additional sources of risk,
it is appropriate to use p as I have defined it. Therefore, I shall
assume that x and L are so stated.
Feldblum will probably have one final point of disagreement:
Covariance and ruin theory approaches usually do not reflect
these additional sources of risk, so to claim that x and L consider
these factors alters his initial assumptions to represent a situation
much more like CAPM, making this an unfair comparison. Further, adjusting loss distributions to reflect these risk sources is
84
RISK LOADS
FOR INSIJRERS
non-trivial, and probably non-objective as well. These are both
reasonable points, but for most lines of business (after adjusting
for special features like retrospective rating), why is the loss distribution not a reasonable first approximation for measuring the
variance of profits? With the exception of parameter risk, most
of the other components have very low variance compared to
losses. Parameter variance must be included, but it is not clear
that measuring calendar year variance of profits directly is the
best way to do this.
Returning to the analysis, consider the following return on
equity equation, which is a form of the equation used by Ferrari
PI:
Rm= Rf + (P/(W))R,,
w9
where Rm is the target return on equity for risk x (and, in fact,
for all risks), Rf is the risk-free rate of return obtained from the
supporting surplus, R, is the return on premium, and P/(zSp) is
the premium-to-surplus (or leverage) ratio appropriate for x. But
wait-shouldn’t
this last statement be a question? Is P/(zSp) an
appropriate leverage ratio for x? The answer is yes, if P/(zS) is
the appropriate leverage ratio for the industry portfolio. To see
this, consider the standard CAPM equation:
Re = Rf + p(R,n - Rf),
(2.10)
where Rm is the return on equity for the industry portfolio. To
solve for the return on premium, R,, needed to produce Ret
according to Feldblum, we subtract the risk-free rate and divide
by the portfolio leverage ratio to obtain:
R, = (Rc- Rf)I(PIW)
= (Rm- R/)I(WWW
(2.11)
But Equation 2.11 is also equivalent to Equation 2.9 if R,,, in
Equations 2.9 and 2.10 means the same thing. Note that Equations 2.9 and 2.10 are two different approaches to the same question: How do we determine the needed profit load? Under Equation 2.10, the CAPM approach, each risk requires a different rate
RISK LOADS
FOR INSURERS
85
of return, but a common leverage ratio is used for all lines. Under Equation 2.9, which I will call the leverage ratio approach, we
target a common rate of return but vary the leverage requirements; in effect, all lines are scaled to the market return.
This last result might seem a little odd-aren’t
the measures
of actual ROE by line of business different under the two approaches? The answer is yes, but this is because the approaches
use different leverage ratios. Under CAPM, the industry leverage ratio P/(zS) is used for all lines of business, whereas under
the leverage ratio approach, the leverage varies by line: P/(zS,D).
If you accept a single leverage ratio, then you must demand differing rates of return on equity based upon the line’s beta. But
if you adjust the surplus requirements hy beta, then you can
accept an equal rate of return on equity across all lines, and
this will equal the industry rate. In practice, the distinction is
largely academic-regardless
of which formula we use, the resulting profit load is the same. Thus, the meanings of R, in
Equations 2.9 and 2.10 are equivalent, and we have our third
important result:
3. CAPM and the leverage approach, which are based upon
the covariance method, are equivalent for computing insurance profit loads.
We also obtain an additional result, a counterargument to
those who suggest that surplus shouldn’t be allocated to line of
business for pricing purposes. If we allocate surplus in proportion to a line’s beta, we obtain a profit load rule that is equivalent to that produced by CAPM. We also normalize the by-line
ROES towards the industry average, allowing the convenience of
targeting a single ROE for all lines instead of varying the ROE
target by line. You can obtain the same results by not allocating
surplus, using the industry leverage for all lines and varying the
target ROE according to CAPM. I agree with Feldblum that this
allocation of surplus has nothing to do with solvency considerations, but that it produces a pricing rule that makes economic
86
RISK I.OADS
FOR INSIJRERS
sense, This allocation of surplus differs from Gogol’s method in
that it is simply a “grossing up” of the marginal surplus requirements so that they balance back to the total.
But now we have a problem. In creating this nice link to
CAPM, we seem to have lost our way back to the original
ruin theory equation. According to CAPM and the leverage
approach, the appropriate surplus for the industry is zS, but
Equation 2.1 says this value should be ZS - R. R is not small
compared to zS, and there is no adjustment that brings these
approaches into line. However, let us recall Philbrick’s concern for “the overly simplistic binary division of the world into
solvent and insolvent companies. Gradations of insolvency are
important.. . .” We could reflect this by adopting a more aggressive ruin constraint: for example, that the sum of loss and expense. minus profit, may not exceed premium plus available surplus (i.e., just surviving is not good enough). In this case, the
needed surplus would now be zS, as per CAPM and the leverage
approach. This is no longer strictly ruin theory, but it is certainly
related, and our analysis provides evidence that this is a more
“financially sound” approach than pure ruin theory.
That leaves us with the two final approaches, utility theory
and the reinsurance method. As Feldblum noted in his paper,
neither of these approaches has straightforward equations with
which to work, so this part of the analysis will be less rigorous,
and more brief!
The reinsurance approach is not really an independent method,
but is, as Philbrick pointed out, “a powerful reality check.” Presumably, reinsurers are subject to the same market forces as primary insurers, and assuming that marginal risk methods are correct for primary insurers, they should work for reinsurers also.
In fact, this was the context in which Kreps presented his findings. Further, the “reality checking” feature should help both
primary insurers and reinsurers calibrate their estimates from
other approaches and verify that they make sense in the context
RISK l.OAIX
FOR INSIJRERS
87
of the larger market. But this is the extent of this “method’s”
usefulness-it
cannot determine risk loads from scratch, and it
is not the only way one can check the market viability of other
methods.
Utility theory is a more complicated issue. Feldblum notes
correctly that there is no good method for determining exactly
what utility function should be used for determining investor
preferences (or insurer risk loads). However, CAPM requires
as an assumpion
that investors have utility functions of a certain form-specifically,
risk-averse functions with known first
and second moments [lo]. Suppose we consider all of the assumptions required by CAPM except for the utility requirement,
and furthermore, suppose we assume that investors will value risk
as per the CAPM formula. The question is: what does this say
about investor utility? Clearly, it still implies that investors are
risk-averse, because they demand higher returns for taking on
more risk. Do we really need to know anything else? CAPM can
price the risk loads, so why do we need a corresponding utility
method to do the same?
One might argue that we could better price the risk if we
knew more specifics about the market utility function, but this
seems equivalent to knowing how the market rewards different levels of risk, at least for fairly “well behaved” utility functions. One could certainly conceive of investor utility functions
so complex that CAPM no longer applies, but such functions
could probably be shown to fit into the framework of something
like the arbitrage pricing model (APM), which is a generalization
of CAPM [lo]. This is not a trivial step-APM
is considerably
more complex than CAPM, in that it allows investors to use information other than mean and variance statistics to price risk.
This strikes me as an important insight-utility
theory at least
holds out the potential of using more information than just the
first two moments of a portfolio’s probability distribution to determine investor preferences. The following chart describes this
potential shift in approach:
Current
Melhods
Future Metho&
moment-based
CAPM
simple utility functions
simple ruin theory
+
i
+
-+
moments plus other data
APM
complex utility function
complex ruin theory
As used here, “simple” basically means “tractable and understandable.” Most people involved in the field of financial research know that CAPM and related approaches are approximations (hopefully good ones) of a more complex reality. But
tractability becomes less of an issue every year as computing
power increases and research progresses. Understandability is a
more serious issue and may slow progress more than tractability.
I don’t want to pursue this direction any further in this
paper-the
subject would fill a book. As for a verdict on utility theory: 1) for a fairly large class of tractable utility functions,
there is consistency with CAPM and related methods, so it seems
unimportant that we don’t actually have a method to determine
what utility functions to use; and 2) even if more complex utility
functions might model market preferences more accurately, there
are probably other equivalent methods, like APM, that would be
used in practice.
In summary, it seems clear that the five approaches have more
in common than Feldblum, or even Philbrick, would admit. The
key is to carefully state the initial assumptions and eliminate
the various shortcuts and approximations that are so often used
with these approaches. If actuaries continue to ignore covariance
considerations when setting risk loads, for example, then the approaches will not agree, and many of Feldblum’s and Philbrick’s
criticisms will be completely justified.
3. REMAINING
PROBLEMS
This analysis seemingly produces a good result, in that we
now have a single approach for setting profit loads that is ob-
RISK LOADS
FOR INSURERS
89
jective, agrees with financial theory, and could be used in practice. However, the conclusion I have reached is that ~lul~e of the
five approaches deals with some very important and practical
considerations, and without a resolution to these problems, we
end up with risk loads that are driven largely by subjective considerations. This conclusion does not imply that these methods
are unusable-I
use a form of the leverage ratio approach in
practice-but
Feldblum’s article might leave one with the impression that CAPM solves more subjectivity problems than it
actually does. There are many problems that require further attention, but the following are examples that loom large in my
mind.
What is the Industry
Leverage Ratio.?
This is a very important question that CAPM does not answer
and that Feldblum appears to have overlooked. In fact, Feldblum
seems to imply in his paper that once you have computed your
return-on-premium betas, you need only use the “Kenney Rule”
(2-to-1 premium-to-surplus ratio) to convert CAPM return on
equity targets to profit loads! The exact leverage value is not
important-the
point is that Feldblum seems to be saying that
P/(zS) (using the above notation) is known for the industry,
when most certainly it is not. ZS is definitely not statutory surplus, nor even GAAP equity, because these accounting measures
don’t use components that are stated economically (e.g., reserves
aren’t discounted), and we cannot rely upon any given year-end
snapshot of equity to be free of distortions and random fluctuations. Even if we came up with a way to measure S properly
(does risk-based capital do this?-1 have my doubts), what is the
correct value for z? The answer must be something like “whatever the market says z should be,” but this doesn’t help us to
compute a value for 2.
No, there is only one answer to this question at the present
time: P/(zS) must be selected, giving due consideration to the
amount of risk the market and company senior management are
90
RISK I.OAI)S
FOR lNSIJKF.RS
willing to bear (as correctly discerned by Kreps). Once this key
leverage ratio is selected, the other calculations become possible, and it is key because it impacts the profit load for every
individual line of business. So the most CAPM can accomplish
is to compute profit load relati~ities, which is no better than
ISO’s approach, old or new. Perhaps this is what Glenn Meyers means when he says that CAPM requires an allocation of
surplus. Strictly speaking, CAPM does not require one to know
leverage ratios by line, as that is what it computes. but CAPM
most certainly does require that one know the overall leverage
ratio or, equivalently, the leverage for one line of business. It
would certainly be worthwhile to try to develop ways to evaluate
the choice of overall leverage ratio and its accompanying return
on equity (apart from obvious ad hoc methods like comparisons
to industry figures, or other industries with similar risk characteristics. etc.), but that is a subject worthy of a paper of its
own.
Why industry
Leverage Over Company Leverage?
In my analysis, I specifically assumed that the existing large
portfolio was the industry portfolio, rather than an individual
company portfolio as specified by Feldblum. This difference in
assumption does not affect the conclusions of my analysis per se,
but it could produce different risk loads. Indeed, Feidblum notes
that a “small- or moderate-size insurer needs a slightly larger risk
load than that indicated by the industry-wide experience,” in order to pick up some of the specific risk. 1 question this: why
would an insured be willing to pay this additional charge? One
could argue that a small insurer may be less “solid” than a large
insurer because the small insurer is more affected by random
fluctuations in experience. The risk of insolvency is higher and
thus the small insurer offers a “lower quality product” and thus
demands a lower premium. This is a simplistic argument with
problems of its own, of course, but I have heard it made. Although I agree that an insurer may possess additional risks versus
RISK LOADS
FOR INSURERS
91
other companies, I don’t see why an insured would pay for this
difference.
Why shouldn’t “equivalent” lines of business demand equal
risk premiums in a competitive market? This question is almost
tautological. The answer that “every insurer is different” might
be a hard sell to insureds, particularly less-sophisticated insureds
(e.g., as in personal lines, where the products are relatively simple risk-transfer mechanisms and are largely interchangeable between companies). A riskier insurer needs to do something, probably via reinsurance or a portfolio change, to “steer” its portfolio
towards the market optimal portfolio that is less risky. Using a
“market equilibrium”-type argument, shouldn’t insureds pay only
the competitive equilibrium risk charge for all interacting companies, and doesn’t this mean that beta should therefore be measured against the market return as opposed to a company’s overall return?
Also, do not assume that “market” means the insurance
market-in
view of overall concerns for asset/liability management, why shouldn’t we measure risk against the entire market?
Actually, this is perhaps too big a stretch-in
his paper, Feldblum points out some valid reasons why insurance contracts differ from financial instruments. But surely the risk inherent in
the investment portfolios varies among insurers. The extent to
which the investment portfolio does not interact efficiently with
the insurer’s underwriting book is another risk for which insureds
may not be willing to pay. This line of reasoning starts to touch
on areas outside of underwriting risk. For example, insurers are
exposed to asset risks that are not directly related to their underwriting risk, such as the risks associated with stocks, real estate, or venture capital. From a stockholder’s perspective, these
asset risks are important components of an insurer’s beta; but,
arguably, these forms of investment should pay their own way
and should not be charged back to the policyholder in the form
of a higher risk load. It would seem that only those risks that
arise from the interaction of investment and underwriting that
92
RISK LOADS
FOR INSURERS
cannot otherwise be diversified away should be included in insurance risk loads. Realistic examples of this seem hard to come
by: one could envision a deal to pay an insured a guaranteed
rate of interest on the funds held for a large deductible account,
and the rate might be higher than current Treasury rates. The insurer would certainly have a right to charge for this, but I suspect
that such an arrangement would more likely be struck, with little
consideration for an adequate rate, simply to get the account.
There is another reason why the distinction between industry
and company risk is important. If one measures risk against a
company portfolio only, it is possible that individual transactions
could unduly influence the risk calculation. An example would
be large assumed reinsurance contracts. Although such considerations are important to the insurer, there is still the question
of how much of this cost to pass down to the insured. In his discussion of Kreps’s paper, Gogol correctly identified this issue as
a problem of “order dependence” (i.e., that the risk load changes
depending upon when the risk is written) and developed a formula to correct for this. Similarly, using a larger market base
forces the risk measurement of individual contracts closer to the
margin, which equalizes risk charges and better satisfies CAPM
assumptions.
There are no definitive answers to these questions. The practical effect of these concerns would be to shift an insurer’s total
risk load up or down, equivalent to changing the overall industry leverage ratio, and in practice this value is selected as noted
above. The point is that the CAPM methodology proposed by
Feldblum has not resolved these issues, although it is a very good
framework within which to further discuss the problems.
How to Compute
Covariance?
The fact that CAPM is a theory that applies specifically to
financial securities means that assumptions will be needed to
adapt the approach to measuring insurer risk. For example, Feld-
RISK LOADS
FOR INSURERS
93
blum states in a footnote that CAPM “has obviated the need for
quantifying covariance.” This may be true for stocks, but not for
insurance profit loads. As stated before, empirical profit information is not the best starting point for this calculation. Feldblum
mentioned at least two problems that require attention:
l
l
adjusting for reserve deficiencies and redundancies (i.e., getting to an accident year basis); and
using discounted cash flow to allocate investment income.
There is a third problem, and it’s a big one. To estimate the
by-line betas, we need a series of historical operating ratios by
line and in total in order to perform the required regression.
However, what we want is an estimate of the current beta for
a line. Doesn’t that require our data to be at “current level”?
Moreover, “current level” comprises a lot more here than just
rates and trends-hanges
in mix of limits, legal climate, social
conditions, and the like are much more important in an analysis
of risk than in an analysis of expected cost. Add to this the numerous other calendar year distortions faced by insurance companies, and calendar year data becomes very messy indeed. It
isn’t clear that the most fruitful approach is to start with calendar year data and to expend a lot of effort cleaning it up. Actuaries simply have more troublesome data problems than do stock
analysts in this instance!
The reviewers of this paper brought up a good point. Bringing
data to current level has the effect of reducing the variance of
historical loss ratios that resulted from shifting conditions. but
these shifts reflect legitimate risks to the company and should
be included in the cost of capital. I agree with this to a point.
By bringing data to current level, my hope is to obtain a good
measure of process risk. However, this procedure does eliminate
valid sources ofparameter
risk that somehow must be measured
and included. I prefer to separate the two measures and try to
obtain a clean estimate of each. Further, some of this perceived
risk cannot be passed on to insureds-for example, the risk due
94
RISK I.OADS
FOR INSlJRERS
to a period of deliberate underpricing to gain market share. This
is something that a company inflicts upon itself, and we cannot
expect future policyholders to accept risk loads computed using
past “price volatility.”
Rather than a straight CAPM approach with calendar year
data, begin instead with a model of current accident year losses,
adjusted for all current conditions and including a measure of
parameter risk. The advantage of this approach is that models
of this form probably already exist for pricing and/or reserving
purposes. One problem, of course, is that we must also include a
measure of this distribution’s covariance to the market. In practice, the only source for such information is the same kind of
calendar year industry data used by Feldblum, but such data are
very difficult to work with even for this more narrow purpose,
and don’t produce very “intuitive” results (such as the low beta
for surety computed by Feldblum and noted by Philbrick). In
most cases, it is necessary to ipzore the covariance terms and
to use instead a simplification that is more practical (e.g., one
based upon standard deviation). It is preferable not to do this,
but we must realize that this is an approximation to the correct
answer. Further, we should continue to explore ways to better
measure and incorporate covariance.
It boils down to a choice between simplifying assumptions:
use CAPM with calendar year data adjusted “top-down” as best
you can, or start “bottom-up” with an accident year model and
reflect as many sources of risk as possible. I prefer the latter.
and I presume Feldblum would advocate the former: but both
are approximations and need more research.
4. CONCLUSION
Feldblum and the many other contributors to this subject
should be congratulated and encouraged to continue the discussion. Most of what they have said has significantly advanced the
state of the art in measuring risk loads. My message is directed
RISK LOADS
FOR INSIJRERS
95
primarily to those less familiar with these issues, and that message is 1) the show has just begun, and 2) the show to date has
largely consisted of variations on a common theme. My concerns
are only a sampling of the issues needing resolution-this
topic
should be fertile ground for inquiry for some time to come.
96
RISK LOADS
FOR INSURERS
REFERENCES
[l] Feldblum, Shalom, “Risk Loads
LXXVII, 1990, pp. 160-195.
for
Insurers,”
PCAS
[2] Philbrick, Stephen W., Discussion of Feldblum: “Risk Loads
for Insurers,” PCAS LXXVIII, 1991, pp. 56-63.
[3] Feldblum, Shalom, Author’s Reply to Discussion of Feldblum: “Risk Loads for Insurers,” PCAS LXXX, 1993, pp.
366-379.
[4] Kreps, Rodney, “Reinsurer
plus Requirements,”
Risk Loads from Marginal SurPCAS LXXVII, 1990, pp. 196-203.
[5] Gogo!, Daniel F., Discussion of Kreps: “Reinsurer Risk
Loads from Marginal Surplus Requirements,” PCAS LXXIX,
1992, pp. 362-366.
[6] Meyers, Glenn G., “The Competitive
Risk Load Formula for Increased
PCAS LXXVIII,
1991, pp. 163-200.
Market Equilibrium
Limits Ratemaking,”
[7] Robbin, Ira, Discussion of Meyers: “The Competitive Market Equilibrium Risk Load Formula for Increased Limits
Ratemaking,” PCAS LXXIX, 1992, pp. 367-384.
[8] Meyers, Glenn G., Author’s Reply to Discussion of Meyers:
“The Competitive Market Equilibrium Risk Load Formula
for Increased Limits Ratemaking,” PCAS LXXX, 1993, pp.
396-415.
[9] Ferrari, J. Robert, “The Relationship of Underwriting, Investment, Leverage, and Exposure to Total Return on Owners’ Equity,” PCAS LV, 1968, pp. 295-302.
[lo] D’Arcy, Stephen F?and Neil A. Doherty, The Financial
The-
Property-Liability
Insurance Contracts, Monograph No. 15, Homewood, Illinois: Richard D. Irwin, 1988.
ory of Pricing
AUTHOR’S
REPLY TO DISCUSSION
Abstract
The actuarial theory of insurance risk Loads has foilowed a meandering course. Actuaries have approached
this subject with diRerent perspectives, contributing
important but seemingly unrelated insights. Toad Bat&s
masterful discussion of “Risk Loadr for Insurers” demonstrates the connections
between the different
approaches, thereby laying a firm foundation for a unified
theory.
Bault first shows the consistency among the risk
load procedures proposed by Rodney Kreps, Stephen
Philbrick, and Shalom Feldblum; he concludes with several issues that warrant further analysis. This reply follows a similar format, beginning with the current applications of risk loads and risk margins, and then addressing three of the issues that Bault raises.
1.
RISK LOADS: FROM THEORY
TO PRACTICE
“The show has just begun,” says Bault, and he foresees an
exciting future for this drama. He is correct; indeed, the future
has already begun.
‘Bventy years ago, when Robert Miccolis [4] wrote his seminal
paper, “On the Theory of Increased Limits and Excess of Loss
Pricing,” the subject of risk loads was considered too theoretical
for most actuaries. The practicing actuary was busy determining
persona! automobile or workers compensation rates and relativities. Only a few rating bureau actuaries had the leisure to devise
risk adjustments for increased limits factors.
97
98
RISK LOADS
FOR INSURERS
Now even practicing actuaries deal with risk loads in their day
to day work.
Economic
Reserves
Reserving once consisted simply of determining
mate of an undiscounted indication.
a point esti-
Now companies ask, “What is the economic value of the reserve?” or “What is the true net worth of the company?” Robert
Butsic [l] has argued that the answers to these questions require the consideration of risk margins in the reserves or risk
adjustments to the loss reserve discount rate. Discounting the
reserve at a risk-free rate gives a result lower than the true economic value. The American Academy of Actuaries, in Standard
of Practice Number 20, “Discounting of Property and Casualty
Loss and Loss Adjustment Expense Reserves,” Section 5.5, follows Butsic’s lead:
The actuary should be aware that a discounted reserve
is an inadequate estimate of economic value unless appropriate risk margins are included.’
Quantifying loss reserve risk margins is half the task; the other
half is properly reporting them in financial statements. Stephen
Philbrick [6] has recently proposed accounting procedures for
loss reserve risk margins. If loss reserve discounting becomes accepted accounting practice-as seems likely for property/casualty
companies-the
treatment of risk margins will become a burning
issue.
‘Butsic uses a risk adjustment to the discount rate instead of a risk load in the reserves
themselves. The AAA Standard of Practice No. 20, Paragraph 5.52, considers both methods acceptable: “Explicit margins may be included as an absolute amount and/or through
an explicit adjustment to the selected interest rate(s).”
99
Pricing
When making rates for first-dollar coverage in a line of business characterized by high frequency/low severity losses, such
as personal automobile insurance, there is little need for risk
loads.2 In commercial lines, however, alternative products are
now emerging in which insurers are providing high-risk layers
of coverage. In workers compensation, for instance, deductible
credits on large dollar deductible policies accounted for about
$5 to $6 billion in 1994.3 The risk load is the dominant concern
in the pricing of large dollar deductible policies, since the risk to
the insurer is great yet marketplace competition is severe. The
transition to alternative products in the commercial lines of business has been so rapid that pricing actuaries are now scrambling
to properly estimate risk loads.
Vaha tion
The underlying premise of the NAIC’s new risk-based capital
requirements is that the capital needed by an insurer depends on
the risks faced by that insurer. But how might we quantify the
“risks faced by an insurer”? The quantification of the variance in
the loss estimate, which is the stepping stone for estimating pricing and reserving risk loads, has been extended by Robert Butsic
into the quantification of the “expected policyholder deficit” and
the implied capital requirements [2]. The requisites for an insurer’s “government affairs” actuary were once no more than a
good sense of humor and an endless patience for bureaucracy.
Now the government affairs actuary must understand asset risks,
loss reserve margins, and covariance adjustments (and still retain
the humor and the patience).
?See Wall (71, who shows that the process variance on this business-which
the risk load
is intended to hedge against-is insignificant.
3See NCCI [5, p. 21. In a large dollar deductible policy, the insured
reimburses the insurer for losses up to the deductible amount, which generally is $lGO,OOO
or more. The insurer provides true excess coverage on losses exceeding the deductible
amount. The deductible credit is the difference between the premium for the large dollar
deductible coverage and the premium for corresponding firstdollar coverage.
loo
RISK LOADS
2. THE MARKET
FOR INSIJRERS
AND THE ACTUARY
“What is the industry leverage ratio?” asks Bault. “This is a
very important question that.... Feldblum appears to have overlooked.”
‘Iwo years ago, Feldblum and Butsic were discussing this question. One needs a starting point, a rock to stand upon, from
which all else can be derived. If one knows the appropriate industry leverage ratio for all lines combined, then one can determine leverage ratios for the individual lines of business.
“Tell me the expected policyholder deficit ratio that the company or the industry is comfortable with,” said Butsic, “and I will
tell you the proper leverage ratio.”
“No,” said Feldblum. “Managers and investors are not fluent
in our discourse of risk loads, probability of ruin, or policyholder
deficits. Yet given free markets, they invest funds where returns
are most promising. For pricing purposes, the market is the ultimate arbiter of needed capital, not the actuary. The actuary’s
task is to understand the raw force of the market, not to turn it
back.”
In other words, the existing industry leverage ratio for all lines
combined is our best estimate of the “proper” leverage ratio. If
the leverage ratio is too high, investors will supply more capital.
If it is too low, investors will take their capital elsewhere.4
Of course, not all insurers are equally subject to investor expectations. The capital structures of many mutual insurance companies can be ex lained better by agency theory than by modern
portfolio theory. D Even for stock companies, the judgments of
‘Of course, pricing is not the only determinant of leverage ratios. Regulators may demand
lower leverage ratios for financial solvency purposes. Here the marketplace is at best an
imperfect arbiter.
SAgency theory seeks to interpret the business strategies of company managers, who are
agents of the stockholders or owners. In a mutual insurance company. for instance, will
managers use excess surplus to pay policyholder dividends, incrcasc cmploycc salaries, or
invest in new business operations?
RISK LOADS
101
FOR INSURERS
the capital markets seem inexorably slow. New infusions of capital and the demise of inefficient insurers may stretch out over
decades.
There are cogent arguments for both Butsic’s and Feldblum’s
viewpoints. Butsic is skeptical of the acumen of the marketplace,
and relies more on actuarial expertise. Feldblum is skeptical of
the acumen of actuaries, and relies on the power of the marketplace. Bault comes down on B&sic’s side, though without endorsing any specific procedure. But his basic premise is correct:
This is a central issue in estimating risk loads.
3. INDUSTRY
LEVERAGE
VERSUS COMPANY
LEVERAGE
“In my analysis,” says Bauit, “I specifically assumed that the
existing large portfolio was the industry portfolio, rather than an
individual company portfolio...Although...an
insurer may possess
additional risks versus other companies, I don’t see why an insured would pay for this difference.”
Bault is correct. Pricing begins with the market, whether for
insurance companies or for other firms. Company-specific analysis tells you only whether the prices achievable in the marketplace are adequate for your company. If the actuary says, “Our
firm needs greater returns, so let us raise our premium rates,”
the firm will succeed only in losing market share.
4. QUANTIFYING
THE RISKS
“Some of the risk,” says Bault, “cannot be passed on to
insureds-for
example, a period of deliberate underpricing to
gain market share. This is something that a company inflicts
upon itself, and I don’t see how one can expect future policyholders to accept risk loads computed using past ‘price volatility.“’
Do companies deliberately underprice during underwriting
cycle downturns, perhaps to consciously inflict pain upon them-
102
RISK LOADS
FOR INSURERS
selves? Actuaries are aware of the premium inadequacies during
soft markets. Why don’t they just set higher rates?
Oh, the actuaries recommend higher rates, demonstrate the
pressing need for rate increases, and warn of the dangers of inadequate premiums. (They are a garrulous lot, these actuaries.)
But actuaries can only indicate rates; the marketplace sets the
prices.
Individual companies have little choice. Companies that strive
to keep rates adequate-when
rates plummet about them-end
up with adequate rates and no insureds. Underwriting
cycles
stem from the business strategies of incumbent insurers to maximize long-term profits.6 Intelligent insurers learn to “ride the cycles,” so that they partake in the industry’s profits when times are
good but minimize the losses when times are bad. Premium fluctuations are an unavoidable risk of insurance operations. Risk
loads are needed for them just as they are for random loss fluctuations.
5.
CONCLUSION
As Bault points out, risk loads are becoming a staple of actuarial work, yet many issues are still unresolved. Thanks to his
discussion, however, it should be easier to tackle the remaining
problems.
%ee Feldblum [3] for a comprehensive analysis of the nature and causes of underwriting
cycles.
RISK LOADS
FOR INSIJRERS
103
REFERENCES
[l] Butsic, Robert I?, “Determining the Proper Interest Rate for
Loss Reserve Discounting: An Economic Approach,” Evafuating Insurance Company Liabilities,
Casualty Actuarial Society Discussion Paper Program, 1988, pp. 147-188.
[2] Butsic, Robert IT, “Solvency Measurement for PropertyLiability Risk Based Capital Applications,” Journal of Risk
and Insurance, December 1994.
[3] Feldblum, Sholom, “ Underwriting Cycles and Insurance Solvency,” Insurer Financial Solvency, Casualty Actuarial Society
Discussion Paper Program, 1992, Volume I, pp. 383-438.
[4] Miccolis, Robert S., “On the Theory of Increased Limits and
Excess of Loss Pricing,” PCAS LXIV, 1977, pp. 27-59; Discussion by Sheldon Rosenberg, pp. 60-73.
[5] National Council on Compensation Insurance, Back from the
Brink?: 2995 Issues Report, Boca Raton, FL: National Council on Compensation Insurance, 1995.
[6] Philbrick, Stephen W., ‘Accounting for Risk Margins” (with
an introduction by Paul G. O’Connell), Casualty Actuarial
Society Forum, Spring 1994, Volume I, pp. l-90.
[7] Woll, Richard G., Discussion of Cozzolino and Kleinman: “A
Capacity Management Model Based on Utility Theory,” Pricing, Underwriting
and Managing the Large Risk, Casualty Actuarial Society Discussion Paper Program, 1982, pp. l&24.
DISCUSSION
OF PAPER PUBLISHED
IN VOLUME
LXX11
A SIMULATION
TEST OF PREDICTION ERRORS OF
LOSS RESERVE ESTIMATION TECHNIQUES
BY JAMES N. STANARD
DISCUSSION
1.
BY EDWARD
E PECK
INTRODUCTION
This discussion of James Stanard’s paper “A Simulation Test
of Prediction Errors of Loss Reserve Estimation Techniques” will
use his simulation technique to test three loss reserving methods.
Two of these methods are discussed in Stanard’s paper, and one
is relatively new having been presented in the Proceedings last
year by Daniel Murphy [6]. The three methods are shown to be
special cases of a general weighted average approach. In addition, some of the concepts presented by Stanard concerning the
expected value of a loss development factor will be analyzed in a
little more detail. Please note that the results derived in this discussion are due to the assumptions made within this discussion
and may not be applicable to general loss reserving situations.
2. THREE LOSS RESERVE METHODS
To describe these three methods, the following notation will
be used: if Xi,i represents a random sum of losses from accident
year i, measured j years after the beginning of the accident year,
then an accident year loss triangle is as shown in Table 1.
An age-to-age average loss development factor from age j to
age j + 1 can be defined as
LDFj = CLx,,j+l/X,jI/n!
104
106
PREDICTION
3.
ERRORS
OF LOSS RI:SI~RVl:
WEIGHTED
I:S’I‘lMA’IlON
AVERAGE
‘l’fi(‘I INIOIJf:.S
APPROACH
Suppose the observed value Xi,j is regarded as a “fixed” or
controllable value and is used to predict the random value Xi,,+l.
Since X;,j is not considered a random variable it will be written
in lowercase as X;,j. To estimate Xi.;+,. it would make sense to
use a weighted average of the available s;.js. The weights are
given as
Wi,j
I
1
= Xi,j
-xi,, 3
/x
i
where Ci W;*j = 1.0. An age-to-age link ratio is then given by
LDF; = 1 w,;X.,+dx,,,
The three methods described in Section 2 can be viewed as
special cases of this general weighted average. Table 2 relates the
methods and weights.
TABLE 2
WEIGHTS
USED
If the statistics X,,j+l/.r;+, are from the same distribution (or
different distributions with the same mean), then the weighted
averages will be unbiased since the weights sum to one. This may
not hold for X;,j+l /X,-j. where the denominator is viewed as a
random variable, as will be discussed later.
Assuming for the time being that .Y,,, is fixed. it could be
helpful to consider the variance of X,.j+, in deciding which set
of weights to use. In some cases. the variance of X,,i+, for a
given xi,j may depend on the size of -vi,,. For example, a “large”
PREDICTION
ERRORS
OF LOSS RESERVE
ESTIMATION
TECHNIQUES
107
value of xi,; could typically be followed by a small variance in
xi,j+l*
If the Xi,i+lS are independent and their variances for a given
xi,j are given by sfj+ly then define the random variable
and
i
Var(Kj)
(3.1)
= C Wtj/&s$+l.
If the variance of Xi,j+i for a given xi,j depends on the size
of xi,j, one possible way to relate the two is to consider S:j+l to
be proportional to x~,J:
2
si,j+l
r
Ci xi,j.
(3.2)
Note that r < 0 is possible and would imply an inverse relationship between the size of loss and the subsequent variance.
Substituting
vields
,
the right side of Equation
Var( Kj) CCC X&*d,j*
3.2 in Equation
3.1
P-3)
The variance of Kj as a function of Xi,j is developed here to help
choose weights and therefore a reserving method. As Stanard
points out, an estimator should be unbiased and have a minimum
variance.
It can be shown (see the appendix) that the weight structure
that minimizes the variance of Kj is
Wij
2-r
= Xif
x2:’
l2J ’
c
i
This leads to choosing the usual arithmetic averages (Method I)
if r = 2, Method II if r = 1, and Method III, the least squares
estimator, if r = 0.
108
PREDICTION
ERRORS
OF LOSS RESERVE
ESTIMATION
TECHNiQUES
Applying all of this to a loss development triangle, the question is whether the variance of the sum of losses at a particular
point in time is dependent on a previous measure of losses. One
way to check differing variances at various levels of a predictor
variable xi,j is to plot the residuals. Unfortunately, there aren’t
enough points to look at in most loss reserving situations even if
a consistent relationship between accident years is assumed. In
some cases, however, one may believe that greater early development of losses commonly reduces the variance of the next period
loss level. If this is the case, it would make sense to choose r less
than zero.
Exhibit 1 displays the results of applying the Methods I, II,
and III using the simulation procedure outlined by Stanard. Recall that Methods I, II, and III correspond to r values of 2, 1,
and 0, respectively, depending on the variance assumption. Also
tested are weighting schemes where r is set equal to - 1 and -2.
This would correspond to the case where there is an inverse relationship between the variance and the previous size of loss as
discussed above. It is interesting to note that the mean prediction
error decreases as r decreases.
These results show that r = 0 (Method III) produces the
smallest prediction error for the current accident year, but the
prediction of previous accident years can be improved by using r less than zero. Given a knowledge of the underlying structure of loss development, as is the case in this simulation
model, it would be possible to choose an optimal value of r
for the specific structure. In fact, r doesn’t have to be restricted to integers; it could take on any real value and even vary
by accident year. Finding an optimal r would be nearly impossible with actual loss data due to the lack of sufficient data
and changes in underlying reporting patterns. But it could be
possible to find a range of r values that would improve estimates.
PREDICTION
ERRORS
OF LOSS RESERVE
4. AGE-TO-AGE
ESTIMATION
TECHNIQUES
FACTORS-LOG-NORMAL
109
MODEL
If we regard xi,j more realistically as an observation of a random sum Xi,j at time j, followed next period by loss Xi,j+i, then
pairing them, (Xi,j, Xi,j+r), adds another dimension to evaluating
their relationship.
Stanard points out in his appendix that, in general,
W/Xl f WI/WV
In the case of losses emerging and or developing and the notation used here,
E[Xi,jtl/Xi,j] # E[Xi,jtl]/E[Xi,jl.
So, using the average of development factors to develop ultimate
losses could lead to incorrect conclusions.
For ease of presentation, the random variables Xi,j and Xi,j+i
will be represented by Xl and Yl, respectively, from here on in
this section. Using this notation, the issue is, what is the expected
value of the statistic Z1 = Y1/X1? To investigate Z1, the pair of
losses (X1,Yl) will be modeled as an element of the joint bivariate log-normal distribution where Xr and Yl are possibly related
via a correlation coefficient. Other joint distributions may be appropriate, and the choice depends on the characteristics of the
data in question. The log-normal leads to very convenient computations, as will be seen.
If and only if X1 and Y1 are jointly log-normal, then X =
In(Xl), and Y = In(Y1) would be joint normal variables. In this
case. a loss development factor is given by the statistic
Zl
=
Yl/Xl
= exp(Y)lexp(X)
= exp(Y - X).
110
I’HiiDICI-fON
fiRKOI<S
01’ I.OSS f<l:Sf:lWl
1:S’f’IMAIION
‘I I:(‘1 INfOlJI:S
This form is convenient due to the fact that the expected value of
Z1 is easy to find using the moment generating function of the
bivariate normal. M(tl, tz) will denote the moment generating
function of the bivariate normal with the following parameters:
pI1-= mean of X,
pLv= mean of Y.
err = standard deviation of X.
U)’ = standard deviation of Y, and
p = correlation
where X = In(X,)
coefficient of X and Y.
and Y = In(Y1).
M(t,,12) = exp[rIp,
+ tips + ($1~: + 2f1f,f2g~u,~ + ri4)/2]
E[Zl] = EIYI/XI]
= E[exp(Y - X)]
= M(-1,l)
= exp[py - p.K + (d - +a,+
+ 4)/2].
Since
E[XII = M(1)
= exp(pL, + at/2),
and
W’II = M(1)
= exp(py+ n.t/%
then
EIYl]/E[&]
= exp[py - pLx+ (fl: - d/21.
PREDICTION
ERRORS
OF LOSS RESERVE
ES’IlMATlON
111
TECHNIQUES
Getting back to the question of whether E[Yl/Xr]
EIXI], define the ratio
# E[Yr]/
d = E[Y~/X~l/[E[Y~l/E[X~lI
= exp[py- px+ (4 - 2pr~~ + 4)/4
exp[py- pL*+ (0; - d>/23
= exp(az - pa,gy).
But
where
crxy is the covariance of X and Y.
d = exp(az - gxy).
This ratio d is the theoretical ratio of the expected straight average LDFs to the expected weighted average LDFs. Note that d
is greater than 1.0 when CZ > uxy and E[Yl/Xt] > E[Yl]/E[Xl].
To investigate d, the following simple model of loss development similar to Stanard’s is created. Assume:
1. Losses from a Pareto severity:
F(x) = 1 - (15,000/(15,000 + x))3;
2. A normal frequency (mean = 50, variance = 25);
3. An exponential reporting pattern:
P(n) = 1 - exp(0.75n);
4. Five “periods” are produced (so if the report time is
greater than 5 it is not in the data); and
5. 1,000 samples are produced.
112
PREDICTION ERRORS OF LOSS RESERVti tSI1MATION
TI;CllNIQIJES
The parameters of the log-normal can be estimated from the
sample data using the moments of the transformed variable
ln(Xi). For example,
mx = c ln(X,;) z P,~,
and
i
sx=
[C ln(X1i)2 - m:] II2 z5 ~7~.
i
Some statistics of the log transformed sample data by age of
development are shown in Table 3. The correlations and covariantes are between ages one and two, two and three, etc.
TABLE 3
LOGTRANSFORMED SAMPLE DATA
Age
Mean
Variance
Skew
Correlation
Covariance
1
2
3
4
5
12.12740
0.125657
-0.11204
0.811968
0.084777
12.54222
0.086754
xm3357
0.918299
0.072555
12.68834
0.071959
-0.03813
0.970066
0.067079
12.74793
0.066450
-0.01074
0.980749
0.063972
12.77808
0.064028
0.032122
The next step is to calculate average loss development factors based on the 10s~ data. These would be ci[Yii/Xii]
for
straight average (Method I) LDFs and my/mx for weighted average (Method II) LDFs. Four average LDFs are available linking
each period:
Age-to-Age
Straight Average
Weighted Average
1-2
2-3
.%I
4-5
1S49429
1.485233
1.165989
1.149049
1.063784
1.058583
1.032007
1.029500
Now, according to the d ratio, the ratio of the straight average to weighted average LDFs from the sample data should be
113
PREDICTION ERRORS OF LOSS RESERVE ESTIMATION TECHNIQUES
approximately
d = exp(gf - gxY)
if the distributions are approximately jointly log-normal. The various values turn out to be:
Age-to-Age
Ratio
d
1-2
2-3
34
4-5
1.043223
1.041727
1.014742
1.014299
1.004913
l.cKl4891
1.002434
1.002481
where, for example, Ratio l-2 is 1.043223 = 1.549429/1.485233
and d for l-2 is 1.041727 = exp(0.125657 - 0.084777).
Since the theoretical values and the “experimental” values are
so close, it is worth the effort to check the distributions of the
simulated losses at each period. The Kolmogorov-Smirnov
or
K-S statistic is helpful in measuring the “closeness” of an empirical distribution to a continuous assumed distribution. The hypothesis Ho would be that the sampled distributions are normal
after the In(Xl) transformation. The statistic
Max[]F(x)
- &(x)(]n”2
> 1.36
is significant at the 95% level, where n is the number of data
points. A high value indicates a poor fit and rejection of HO.
For the standardized log transformed data:
Age
K-S
Maximum Difference
1
2
3
4
5
0.5492
0.0174
0.4245
0.0134
0.7075
0.0224
0.6094
0.0193
0.5209
0.0164
The distributions of the standardized log transformed sums of
Pareto variables by period are apparently very closely approximated by a standard normal distribution, and the joint lognormal assumption appears to be valid.
114
PREDICTION ERRORS OF I.OSS RESERW IXIMATION
‘l’l:(I~NIOIJES
The following were calculated using untransformed
ized data from the sample:
Age
KS
Maximum Difference
standard-
1
2
3
4
5
2.1892
0.0693
1.6347
0.0517
1.7544
0.0555
1.4142
0.0447
1xx33
0.0526
These data indicate that a bivariate normal assumption would
not be appropriate for this data.
Concluding this section? the answer to the question “What is
the expected value of an LDF?” is that it depends on the joint
distribution of the losses. The joint log-normal allowed for the
determination of expected LDFs in terms of the parameters of
the underlying variables. It would be possible to use a similar
analysis on actual loss data if reasonable estimates of the distributions of losses by age could be found. Also, this analysis could
be extended to the product of LDFs.
5. SUMMARY
Exhibit 1 displays the results of the three loss development
methods given in Section 2 using Stanard’s simulation routine.
Methods II and III are clearly superior in terms of both bias
and variance. To the extent that actual loss development patterns
are like those simulated, Methods II and III would be preferred
over Method I. As noted above, other weighting schemes may
produce even better results.
Method I, the straight averaging of LDFs, shows the greatest
positive bias. Part of this bias could be explained by the analysis of E[Y/X] in Section 4. An obvious conclusion is that straight
average LDFs will overstate projected ultimate losses, at least according to these models. However, if a selection criterion is used,
such as excluding the high and low LDFs or judgment based on
115
other information, the straight average LDFs would likely produce better results in terms of average error. The goal of the
discussion here is to determine general underlying characteristics of LDFs and age-to-age methods that could possibly have a
bearing on decision making.
The idea of correlation between random sums measured at
successive points in time could give more insight into the selection of loss development factors and age-to-age factor methods
in general. An understanding of how the aggregate distribution
of losses changes with time would be a valuable tool.
116
REFERENCES
[l] Degroot, Morris H., Probability
Wesley, 1986, pp. 300-304.
and
Statistics,
[2] Hogg, Robert and Allen Craig, Introduction
Statistics, Macmillan, 1978, p. 180.
[3] Kleinbaum,
Applied
David, Lawrence
Regression Analysis
Krupper
Addison-
to Mathematical
and Keith Muller,
and Oth.er Multivariuble
Method.y,
PWS-KENT, 1988, pp. 84-85.
[4] Mood, Alexander, Franklin Graybill and Duane Boes, Introduction to the Theory of Statistics, McGraw-Hill, Third Edition, 1974, p. 164.
[5] Mosteller,
Probability
Frederick,
with
Robert Rourke and George Thomas,
Statistical
Applications,
Addison-Wesley,
1970, pp. 4oa-401.
[6] Murphy, Daniel M., “Unbiased Loss Development
PCAS LXXXI, 1994, pp. 154-222.
Factors,”
PREDICTION
ERRORS
OF LOSS RESERVE
EXHIBIT
ESTIMATION
117
TECHNIQUES
1
RESULTSOFLOSSDEVELOPMENTMETHODS
MEAN
Method
r
I
2
1
0
-1
-2
II
III
STANDARD
Method
r
I
II
III
2
1
0
-1
-2
PREDICTION
ERROR
Accident Year
1
2
13,627
13,627
13,627
13,627
13,627
31,498
21,887
16,397
13,532
11,958
DEVIATION
OF PREDICTION
Accident Ear
1
2
170,234
170,234
170,234
170,234
170,234
285,556
278,987
277,716
277,909
278,408
3
4
83,862
40,185
17,110
5,883
67
482,307
121,218
13,056
- 26,840
- 44,583
ERROR
3
4
391,868
347,260
345,466
353,319
363,256
2,406,638
857,74 1
672,590
613,091
592,641
PREDI<Z’llON ERRORS 01; I.OSS Kl:Sl:KVI: 1:s I IMAI’ION ‘I‘lI(‘I 1NIOUtIS
118
APPENDIX
The subscript j will not be used in the appendix for clarity.
The goal here is to find w, such that Var(K) is minimized. If
Var(K) = h(r) = C x[~‘w;?,
where
and
g’(t) = 2 c xf 1 xfln(x,).
i
i
h’(t) = (R’f - f’R)/d
Since g2 > 0, we need to find t to set the numerator equal to 0
or ~‘f = f’~. With some factoring this reduces to
C 1: C .X~+r-2111(Xi) = C xf’+rp2 C xfln(xj).
i
i
i
i
By inspection, t = 2 -
r
solves this equation.
Using the first derivative test, it will be shown that, as t passes
through 2 - r, the sign of h’(r) changes from negative to positive,
indicating that this is a minimum. That is, show
1. Ift<2-r
xxi
i
then
c
x~‘tr-21n(xi)
i
< c J$+‘-’
i
and h’(l) is negative.
2. If t > 2 - r then h’(f) is positive.
c
x:ln(xi)
i
(A.l)
119
PREDICTION ERRORS OF I.OSS RESERVE ESTiMATlON TECHNIQUES
First, let
t<2-r.
Then
2t<2-ri-t,
and
2t+r-2<t.
Also, let
xi > 1.0
i, and xi # xj for at least one (i, j).
for all
These two conditions
considered. Since
are easily met for the loss data being
t>2t+r-2,
and
c x: > c x;t+r-2.
i
i
Equation A.1 is equivalent to the inequality
Cx?f+r-2
1
lIl(Xi)
i
I
< C
Xytrp2
*:lIl(Xi)/
i
i
C
Xi.
i
(A.2)
For given xis, the left side is in the form of a weighted average
of ln(xi) with weights equal to
X2t+r-2
1
C
I
Xft+r-2,
(A.3)
i
and the right side is also a weighted average of ln(xi)
weights
x: xx:.
I i
with
120
So, if
C xDn(xi)/
)Jxt)
(A4
i
i
is a monotonically increasing function of t, Equation A.2 will be
satisfied because 2t + r - 2 < t.
Taking the first derivative of Equation A.4 with respect to t
yields
CX:lIl(Xi)2/cX:
i
i
(JJXh(Xi,)2/
i
(xx:)‘.
i
(A.5)
The form of Equation A.5 is algebraically identical to the variance formula
Var = E[X2] - E[X12,
where the probabilities are the right side weights and the random
variable is ln(xi).
According to Mood, Graybill and Boes [4], the Jensen inequality says that if X is a random variable with mean E[X],
and g(x) is a convex function, then E[g(x)] 1 g(E[X]). It follows that this will hold for Equation AS. In this case g(x) = x2
is convex, so the derivative in Equation A.5 is greater than or
equal to zero. In fact, the only case where the derivative equals
zero is when the probability of a given X is concentrated at a single point, or in this case xi = xj for all (i,j), which isn’t allowed.
This implies that the derivative is strictly positive and Equation
A.4 is monotonically increasing which, in turn, implies that Equation A.1 and Equation A.2 hold since 2t + r - 2 < t. This means
that h’(t) is negative for t < 2 - r, which is what we meant to
show.
If we now consider condition 2 from above, the same argument holds for t > 2 - r, implying that h’(t) is positive. This
shows that h’(t) changes sign from negative to positive, and that
t=2-risaminimum.
ADDRESS TO NEW MEMBERS-NOVEMBER
MICHAEL
13,1995
A. WALTERS
When Allan Kaufman asked me to speak at this meeting, I
was both surprised and relieved. I was surprised that my turn to
give the commencement address to new members had arrived so
soon. About ten years ago, we introduced the tradition of having a prior generation past president come back one more time
to give such an address at each CAS meeting. With two meetings a year, we now appear to be running out of past presidents.
Perhaps we should admit new members only once a year.
I was also happy (and relieved) that I was not elected ten
years from now. By then, unless the tradition changes, new
presidents-elect will be asked to give a past president’s address
at the same meeting they are elected president. Even worse, ten
years later, one of the new Fellows may have to be chosen at
random to give the commencement address as a possible future
past president.
Now, thanks to the 100 new Fellows this time, the CAS has
allotted only nine minutes for these remarks-nine
minutes to
congratulate you for your accomplishments and to inspire you to
be good CAS corporate citizens in the future. You can do the
latter by volunteering for professional activities and by wearing
your new credentials with pride and with gratitude toward the
CAS.
And you do deserve congratulations for the eight or more
years of hard labor spent in passing the exams. No doubt, most
of you studied more for the CAS exams than you did collectively
for the 40 or so final exams in your eight semesters of college.
The good news is that this phase of your life is over. You can
now catch up on lost reading or renew some neglected hobbies.
At the same time, don’t get too distracted from your professional interests, because the bad news is that you probably have
121
122
ADDRESS ‘l-0 NEW MEMBEKS
as much continuing education ahead of you as you had basic education to become an actuary.
The fact is that the world of casualty risk will undoubtedly
change much more rapidly in the next 25 years than it did over
the last quarter of a century, and the changes of the past 25 years
occurred at warp speed compared to those of the previous 25.
What lies ahead for the actuarial world is a staggering amount of
information generated by computers and the obligation to use it,
measure it, and make recommendations on how to deal with it.
Actuarial opinions on loss reserves are just the tip of the iceberg. Dynamic financial analysis (DFA) is in its infancy, but the
ramifications will be enormous. Today’s reserve opinions are really for only a piece of one side of the balance sheet. It will be
up to the appointed actuary of the future to articulate and measure the potential variations in future results from plausible risks
to the surplus of the entire insurance enterprise. Plus, you will
get to use all aspects of your training: pricing, reserving, finance,
assets, and valuation.
DFA evolved as the original response of the actuarial profession to the flurry of large insurer insolvencies around 1990. That
response was a variation of the British system where regulators
required solvency tests performed by appointed actuaries. This
appeared to be where Representative John Dingell (D-MI) was
headed when he was calling the shots in the U.S. Congress.
But our profession has backed away from that initial trial
balloon, and is now pursuing dynamic financial analysis reports
as valuable in their own right to all companies. The profession
doesn’t even need to require them. When DFA reports become
routine and cost-efficient, they will effectively be mandated for
all insurers through their appointed actuaries. This is because
failure to perform DFA tests and communicate problems to senior management would constitute dereliction of duty by the appointed actuary. Thus, dynamic financial analysis may become a
reality for all appointed actuaries early in the next century.
ADDRESS TO NEW MEMBERS
123
There will undoubtedly be new types of insurance and reinsurance, as portfolios of risks will be packaged-like
mortgages
of the past-into
fungible components that can be transferred
to nontraditional risk bearers around the world. The actuaries
of the future will have to price these transactions and provide
valuations for the balance sheets of the new risk bearers.
It sounds like a formidable task, especially because some of
these mechanisms haven’t been invented yet. But casualty actuaries have staked out their turf in being responsible for keeping
score for the future world of non-life financial risk.
How will you maintain your expertise and professionalism to
deal with the casualty risks of the future? None of the basic descriptive material has even been written, much less assigned to a
syllabus of continuing education. The answer is simple: the same
way you succeeded in passing the exams. You demonstrated the
ability to get the essence of a problem in a short period of time,
to understand complex relationships, to solve the problem, and
to articulate it to an audience. These skills, honed by the actuarial exam process, are the key to handling all new actuarial problems confronting you. In fact, the resiliency skills you’ve demonstrated during this rigorous process are analogous to the ones
that made the football teams coached by the legendary Vince
Lombardi so successful.
For those who don’t remember professional football in the
1960s one team from a small town in Wisconsin dominated the
league. Green Bay was, undoubtedly, the smallest market of any
professional team in the history of U.S. sports. Without the large
market resources to lavish on players and coaches, the team
hired a head coach named Vince Lombardi, whose only previous head coaching experience was at a small high school in New
Jersey-St. Cecelia’s.
But Lombardi had a system of training-physical
and mentalthat built a winning tradition. Even those of us who never experienced a professional football preseason can relate to the concept
124
of a rigorous mental training program. Lombardi’s players hated
the intensity of his training camp and genuinely feared what he
would ask them to do next. Nevertheless, it was Lombardi’s belief that if you survived his training camp, the regular season
would not seem so tough. And, the motivation to win a close
game in the fourth quarter was tremendous. No one wanted the
pain of those preseason sacrifices to have been in vain, so the
efforts expended by his teams in the final quarters were astounding.
Many of you have experienced this in the final month of study
before an exam. If you’d already invested three hundred hours
of study time, you were not about to slack off in the last few
weeks before an exam. That may also explain why the CAS recommends that you keep track of the number of hours you study
for each exam. This same attitude of preserving your previous
effort will carry over to the next phase of your professional
career-not
just in motivating you to tackle future challenges,
but also in maintaining your professional integrity under difficult
circumstances. The presumption is that those who have invested
so much time in achieving professional status would not jeopardize their careers by unprofessional conduct. At the same time,
there are a few checks and balances to see that you don’t forget
the basic commitment to quality.
The one negative in the Lombardi training regimen-it is now
apparent-was the physical travail that a violent sport like football exacts on one’s body. As glamorous as a Super Bowl ring
may seem to some, the punishing physical activity probably shortens lives and definitely makes for an arthritic middle age. Of
course, none of that affects the career of an actuary. This is duly
recognized in the Jobs Rated Almanac, which consistently rates
actuaries first and NFL linemen close to last in terms of desirable career rankings.
The Vince Lombardi analogy also fails when you consider
that no one is yelling at you to make the grade. You are your
125
own coach and have been for at least the last eight years. Also,
you’re not ending a preseason of training. All of you have been
playing regular season games’ for some time; your training has
taken place between games. Nevertheless, you have still put so
much into the effort, you are likely to maintain that discipline for
long into your careers.
You have come a long way to get here. You have even further
to go in the future-so
much to contribute to this profession,
with your own careers as beneficiaries of that contribution. Good
luck in the rest of the games this season and in all future seasons.
But, just remember, you can’t fire your coach if you don’t make
the Super Bowl.
PRESIDENTIAL
ADDRESS-NOVEMBER
ALLAN
13. 1995
M. KAUFMAN
Introduction
In his 1969 CAS Presidential Address, Bill Hazam quoted the
earlier CAS President Dudley Pruitt, who wondered why presidents give addresses. and who ever reads them after they have
been given. Dudley discovered that presidents give addresses because the Bylaws require it. He further discovered that the addresses are read avidly only by the subsequent presidents.
He was right about presidential reading. In the course of my
reading, I discovered that many past presidents have bravely
forecast future developments. 1 also observed that presidential forecasts are no more accurate than loss reserve forecasts.
Nonetheless, my remarks will include my forecasts. Future presidents, reading this address, will have their chance to wonder what
I was smoking when I prepared these remarks. I can only say that
these directions make sense to me at this time.
With that warning in mind, I will discuss two things: the current state of the CAS, and international directions for the CAS.
How Are We Doing-The
Numbers
First, I love numbers, so let me give you a few current and
projected numbers related to the status of the CAS:
1. We reached almost 2,500 members at this meeting2,489 to be precise. We will easily double that in the next
ten years (2005). Those new members are generally already at work in the profession; so we know the new
members’ names, not just the numbers.
2. Over the following ten years (2015). we will grow by
50 percent to 100 percent to between 7,500 and 10,000
members.
126
PRESIDEWIIAL ADDRESS
127
3. We are over 13 percent of the North American actuarial profession. Twenty years ago we were a bit under 12
percent of that group. In the next twenty years, our share
will continue to grow.
4. Worldwide, the CAS is one of the largest actuarial organizations. After the Society of Actuaries (SOA) and the
American Academy of Actuaries (AAA), those “megaorganizations,” we are roughly the same size as the next
set of actuarial organizations; for example, the U.K. Institute and Faculty of Actuaries and the Canadian Institute of Actuaries.
How Are We Doing-Other
Considerations
Besides raw numbers, we can observe how we are viewed by
some of our important audiences-the business, regulatory, and
actuarial communities.
1. We have the respect of the insurance business community. There are jobs for all of us. We are wellcompensated relative to many other insurance professionals. The demand for actuaries has seldom been
higher. When two New York-based insurance companies
effectively discontinued operations and two actuarial departments became unnecessary, there was still essentially
no actuarial unemployment.
2. We have the respect of the regulatory community. That
is evidenced by reserve opinion requirements, the role
played by the actuarial profession in the development of
property/casualty risk-based capital, and the fact that any
regulatory consideration has been given to Dynamic Financial Analysis (DFA) requirements.
3. Within the actuarial profession, we are viewed as a
model for coordination with the AAA and we are sought
after as a partner with the SOA.
128
PRESIDENTIAL ADDRESS
Recent Demonstrations
of Our Strengths
The ability to maintain this strong position depends on our
capabilities in the areas of research and education both basic
and continuing. Let me point to a few recent demonstrations of
those capabilities. First, our DFA research effort has given us
a chance to proactively expand the scope of skills available to
our members. Second, in the last two years, there has been a
major change among insurers in the level of recognition of environmental liability exposures. Roughly four billion dollars have
been added this year alone. While there have been many forces
at work to produce those changes, I do not think that we are
simply flattering ourselves if we acknowledge that the research
work of our members and the publication and seminar efforts
of the CAS have also contributed to that recognition. Third, in
the research area, the use of catastrophe modeling has expanded
enormously. While the profession was not the leading force in
the emergence of these models, we have and increasingly will
become more involved in the application of these tools.
We have been reaching out to academics in our efforts to
build our knowledge base. One of the most exciting efforts is
the CAS-financed DFA Variables project by Assistant Professor
James R. Garvin, Ph.D., of the University of Texas at Austin.
The CAS continues to increase the number of continuing education events, and our members continue to take advantage of
those opportunities. In July, we held the first Dynamic Financial
Analysis seminar. This seminar is intended to be a regular offering; it is expected to grow from its current “special interest”
size of about 150 attendees to a much larger scale, perhaps even
the size of the CLRS and Ratemaking Seminars. The 1995 seminar was held in Atlanta-a
rather warm location for a July seminar. In 1996, this summertime seminar will move to Montreal, a
cooler location.
We continue to explore new topics for classroom-size seminars, and new types of seminars. For example, we held the first
PKESIDENTIAI. ADDRESS
129
participant-led limited attendance seminar on Actuarial and Financial Risk Theory in October in Boston, prior to the Environmental Liability seminar. In that seminar the participants were
the teachers. A description of the seminar is in the November
1995 issue of the Actuarial Review.
Future Strengths
Looking ahead, there are areas where our skills have not
reached the levels that will likely be required in the future. Two
of the most important examples are (1) our knowledge of health
insurance as it affects casualty coverages, and (2) our understanding of the asset portion of the property/casualty company
balance sheet. A newer example relates to an increasing number of efforts to take insurance risk and transform it into capital market risk. It may become more and more difficult to distinguish insurance from investment banking. Finally, we need to
export our insurance-related expertise to areas covering similar
risks that are not considered insurance. That means corporate
risk management in all its varieties. Our members and research
committees recognize these weaknesses, and a number of activities are underway to address them.
Society of Actuaries
A few words on the CAS relationship
order at this point.
with the SOA are in
In many of the cases I just identified, the boundaries between
the “casualty” discipline and the actuarial disciplines included in
the SOA are undefined and perhaps undefinable. This is a fact
of life, which illustrates why our level of coordination with the
SOA in the research area will likely increase.
We have a long-standing and successful involvement with the
SOA in the basic education process. They are again reviewing
their educational structure. They have invited us to assist them
in their review, and we have accepted that invitation. They have
130
PRES1DENTIAI. ADDRESS
recently invited CAS involvement in a number of research areas: the North American Actuarial Journal and the North American Actuarial Foundation that I mentioned in my “From the
President” columns during this year. Those are areas of cooperation that the CAS Board will be assessing in the coming months.
There are issues regarding specific suggestions; but, to me, closer
coordination with the SOA in the area of research, like our existing close relationship in the area of education, is inevitable and
desirable.
While we are always somewhat nervous about cooperation
with our large sister organization, our continued strength means
to me that we should be proceeding with confidence, not concern.
Our Business Has Become Increasingly
International
I want to spend my remaining time discussing another fact
of life and its implications for us: our business is increasingly
international.
One of my responsibilities as president-elect and president
was direct involvement in CAS international relations. In that
role, I participated in the creation of the new IFAA. the Intemational Forum of Actuarial Associations. I attended three annual
meetings of the U.K. General Insurance Study Group, somewhat
like a mini-CAS within the U.K. Institute of Actuaries. Finally. I
had the opportunity to go to Japan last vear and speak on behalf
of the CAS to various actuarial and business groups.
Many of us have also had some international business experience. We have all observed that non-North American companies own many of our employers and clients. In addition, many
of our North American-based employers and clients own nonNorth American insurance operations. While this is not new, the
extent of international involvement has changed hy an order of
magnitude.
131
In the past two years, two of the largest U.S. reinsurers completed major transactions, creating a global scope for operations
that were primarily domestic before that. Other North American companies have been expanding more quietly outside of
North America. Similarly, non-North American insurers-some
Swiss names come to mind, but the Swiss are not alone-have
increased their operations in North America and worldwide.
It is true that the major international insurance and reinsurante companies are now among the largest employers of actuaries worldwide. Those companies have had or will be having
international actuarial meetings as they try to take advantage of
their new international casualty actuarial capability. When our
employers behave internationally, can the CAS activities be far
behind?
To establish a strategy, we must understand the international
actuarial community. Let me identify some key points. About
half of the worldwide actuaries are in North America. Most nonNorth American actuarial organizations are much smaller than
the CAS. No other country has a professional organization that is
divided by specialty as the U.S. is divided between the CAS and
the SOA, or as the U.S. pension actuarial profession is divided
among several organizations. Most countries have no health actuarial discipline. In most countries, the casualty segment of the
profession is small or non-existent. Often there is little or no differentiation between life and casualty professionals.
On the other hand, in the U.K., where the actuarial profession is relatively large compared to countries other than the
U.S., there is a growing casualty group, referred to as the General Insurance Study Group. This sub-group of the U.K. Institute
of Actuaries has about 300 members. Twenty years ago, when
that group was formed, there were only 30 members. The U.K.
Institute has recognized that different actuarial specialties may
have different research and education needs, and it has established “boards” by practice area to allow more specialization in
research and education efforts.
132
PRESIDllNIlAL
ADDRESS
As I see it, a casualty discipline has developed in the U.S.,
Canada, the U.K., and a few other countries where two conditions are met. First, the actuarial community overall is sufficiently large; and, second, the casualty or general insurance market is large enough to support a critical mass of general insurance specialists. These conditions have not yet arisen in many
countries. I believe these conditions will increasingly arise, and
separate actuarial disciplines will arise accordingly. This does not
mean separate organizations, just specialization of activities. This
belief is important in my projection of the future CAS international role.
The bulk of the technical general or casualty insurance issues
are the same from country to country-much
more so than is
true of pension or life insurance issues. This observation is also
important in my projection of the future CAS international role.
The process of qualifying actuaries varies around the world.
Examinations are the standard route of qualification in the
English-speaking world and in parts of Asia. On the other hand,
a university degree is a common form of professional training in
Europe and Mexico. In some countries, the actuarial organization is simply a voluntary association.
The two primary models for examination-based education are
the U.K. Institute of Actuaries program, which includes all actuarial disciplines, and the SOA program, which includes all but
casualty/general insurance material.
Let me forecast a few trends:
1. Casualty or general insurance specialty groups will develop as required by national marketplaces. These actuaries will recognize the commonality of casualty issues
globally, will want access to information and research,
and will want to participate with organizations in other
countries in ongoing research in the general insurance
field. The CAS can fill this need.
133
2. These actuaries will associate primarily with their national accrediting organizations, especially in countries
with existing organizations, so we are not talking about
additional CAS members.
3. Emerging countries are looking for an actuarial professional model for their countries. They are likely to
choose an examination process to supplement university
education. No emerging country is likely to adopt the
U.S. model of separate organizations for different specialties. The separation represents an inefficient use of
their limited resources and is not responsive to their current market needs. It is worth noting that the SOA program is at a disadvantage relative to the U.K. Institute of
Actuaries program, because the SOA program does not
include a general insurance segment.
4. The new IFAA will increasingly be a channel of communication among actuarial organizations.
Why does the CAS care? What are our interests? This can
be considered from the perspectives of individual members, the
CAS as an organization, and our employers. I would list the CAS
interests as follows:
1. We want to be sure that the non-North American employers of actuaries recognize the value of CAS membership. That is intended to enhance our value here in
the U.S. and Canada.
2. We want to help North American members who wish to
practice outside the U.S. and Canada. This applies immediately to nearby countries (Bermuda and Mexico)
and longer term to the rest of world. We particularly
want to avoid rules that preclude work in any country.
I do not expect that there will be large numbers of U.S.
expatriate actuaries, but those who do reach out in that
way are our emissaries to the rest of the insurance world.
134
PRESIDEN’I‘IAI. ADDRESS
3. We want recognition that general insurance is a specialty
of actuarial work that requires specific technical knowledge. With this recognition will come efforts to train actuaries in this specialty. This training will be important
to our employers and clients. as they expand worldwide
and need local actuarial expertise in their casualty insurance businesses. To provide training means working with
non-North American actuarial organizations to help educate actuaries in casualty insurance matters. This is important in establishing the CAS as a brand name with
value, rather than as an example of an anomalous specialty behavior of the U.S. actuarial profession.
4. We want to enhance the “image” of actuaries worldwide.
5. International involvement will be professionally interesting and fun.
The important steps, which must be taken to be sure that our
interests are recognized, are as follows:
1. We need to maintain our preeminent role in the U.S. and
Canadian casualty field. That’s obviously what we need
to do regardless.
2. We need to continue to develop and strengthen high
level contacts between the CAS and non-North American actuarial organizations. We want to communicate
to those non-North American organizations that we can
work with both the overall organizations and with their
casualty insurance specialists, if any. We already have relationships with the U.K. Institute and its General Insurance Study Group. CAS participation in the newly
formed IFAA will help make communication with other
actuarial organizations more routine.
3. We need to cooperate in research and continuing education with general insurance specialty sub-groups of
135
non-North American actuarial organizations. This means
both (1) inviting non-North American help on issues of
North American importance (for example the Loss Reserve Uncertainty Theory of Risk project) and (2) offering to provide CAS assistance to non-North American organizations, through committees or otherwise, on
non-North American issues. These activities need to be
undertaken in cooperation with the existing national actuarial organizations in those countries. In this respect,
we are not trying to establish a CAS in other countries.
4. In cooperation with the SOA, we need to integrate casualty material into a complete North American education program suitable for countries that do not have their
own education program. This degree of cooperation with
our large sister organization in North America requires a
high degree of self-confidence about our strengths. I believe that confidence is warranted. Nonetheless, there are
some risks in this direction, so I don’t suggest it lightly.
Again, I do not picture that we are trying to establish the
CAS as the worldwide general insurance organization. I do not
picture that hoards of U.S. and Canadian casualty actuaries will
spill over the world. I do picture that the CAS will be a recognized and valued brand name around the world, and a role model
for generous cooperative participation in the worldwide actuarial
and insurance communities.
Summary
What do we need to do to take advantage of the opportunities?
1. Do the best we can here in the U.S. and Canada.
2. Work closely with the other actuarial organizations-in
North America and globally-on
casualty .matters.
136
The suggested activities will not require much additional CAS
effort, since a number of them are already underway. It does
require courage to work so closely with the SOA in their international efforts. However, a close working relationship with the
SOA in these areas is both necessary and helpful. It does require
the willingness to work through other organizations, but that is
our tradition here in the U.S., vis-a-vis the AAA.
The resulting cross-fertilization of ideas from around the
world will help our own research and continuing education efforts. At the same time, we will establish and reinforce a worldwide CAS brand name that will also serve our members here in
the U.S. and Canada.
Good-Byes
and Thank Yous
A very traditional part of the Presidential
good-byes and thank yous.
Address is the
First, I want to thank our Executive Director James H. “Tim”
Tinsley. In actuarial fashion, I have a two-by-two matrix of
thanks. In one dimension of that matrix, I want first to thank
him personally for his help and second to thank him on behalf
of the CAS. In the other dimension, the thanks are both to him
personally and to the whole CAS office. The office does a marvelous job on our behalf. The CAS office is like Federal Express,
the fax machine, and voice mail. Now that it exists, it’s hard to
understand how things could have ever functioned without it.
Next, I want to thank the current and past Executive Council and Board Members with whom I have served. They helped
make this experience personally and professionally rewarding for
me.
To my partners and colleagues at Milliman & Robertson, Inc.,
I want to report that I will finally be getting back to more immediately productive activities, and I want to thank them for taking
up enough of the slack during my term of office with the CAS to
allow me the time to serve.
PRt:SID1:N’I’IAI.
ADDRESS
137
Nearly finally, I want to thank my wife Fran who encourages
me in all my efforts and without whose love and support nothing
I do would be worthwhile.
And, finally, I want to thank you, the membership, for this
opportunity to have so much fun while serving you.
MINUTES OF THE 1995 ANNUAL MEETING
November I 2- IS. I 995
HOTEL
DEL CORONADO,
SAN DIEGO.
CALIFORNIA
Sunday, November 12, 1995
The Board of Directors held their regular quarterly meeting
from noon to 5:00 p.m.
Registration was held from 4:00 p.m. to 6:OOp.m.
From 5:30 p.m. to 6:30 p.m., there was a special presentation to
new Associates and their guests. The session included an introduction to the standards of professional conduct and the CAS committee structure.
A welcome reception for all members and guests was held from
6:30 p.m. to 7:30 p.m.
Monday, No\‘cmber 13. 1995
Registration continued from 7:00 a.m. to 8:OOa.m.
CAS President Allan M. Kaufman opened the Business Session
at 8:00 a.m. and recognized special guests Jack Turnquist. President of the American Academy of Actuaries. Wilson Wyatt. Executive Director of the American Academy of Actuaries, and John
O’Connor, Executive Director of the Society of Actuaries.
Paul Braithwaite. David Hafling. and John Kollar announced
the 79 new Associates and the 97 new Fellows. The names of
these individuals follow.
138
MINUTES OF THE 1995 ANNUAL
MEETING
139
NEW FELLOWS
Rhonda K. Aikens
Jean-Luc E. Allard
Kerry F. Allison
William M. Atkinson
Karen F. Ayres
Timothy J. Banick
Philip A. Baum
Gary Blumsohn
Donna D. Brasley
Mark E. Burgess
Mark W. Callahan
Kevin J. Cawley
Ralph M. Cellars
Galina M. Center
Francis D. Cerasoli
Laura R. Claude
Mary L. Corbett
David J. Darby
ReneeHelou Davis
Marie-Julie Demers
Lisa Nan Dennison
John P. Doucette
Paul E. Ericksen
Dianne L. Estrada
Madelyn C. Faggella
Michael A. Falcone
Denise A. Feder
Mary K. Gise
Olivia WackerGiuntini
Bradley J. Gleason
Ronald E. Glenn
Marc C. Grandisson
Anne G. Greenwalt
StevenJ. Groeschen
William D. Hansen
ChristopherL. Harris
Matthew T. Hayden
SuzanneE. Henderson
Anthony Iafrate
Patrick C. Jensen
JanetS. Katz
Tony J. Kellner
Brian Danforth Kemp
DeborahE. Kenyon
Kevin A. Kesby
Michael E Klein
Terry A. Knull
Adam J. Kreuser
David R. Kunze
Blair W. Laddusaw
Paul B. LeStourgeon
Marc-Andre Lefebvre
Aaron S. Levine
GeorgeM. Levine
John J. Lewandowski
Maria Mahon
BarbaraS. Mahoney
LawrenceF. Marcus
Robert D. McCarthy
KathleenA.
McMonigle
StephenJ. Mildenhall
RussellE. Moore
FrancoisMorin
Antoine A. Neghaiwi
John Nissenbaum
Victor A. Njakou
Keith R. Nystrom
Edward F. Peck
WendeA. Pemrick
Mark W. Phillips
JosephW. Pitts
Denis Poirier
On Cheong Poon
Arlie J. Proctor
Mark S. Quigley
Donald A. Riggins
Bradley H. Rowe
John M. Ruane
JamesV. Russell
PeterSenak
Rial R. Simons
Keith R. Spalding
Douglas W. Stang
Richard A. Stock
Marianne Teetsel
Cynthia J. Traczyk
Patrick N. Tures
PeterS. Valentine
CharlesE.
Van Kampen
David B.
Van Koevering
Kenneth R.
Van Laar, Jr.
Mark D. van Zanden
Trent R. Vaughn
Lisa Marie Walsh
John S. Wright
Claude D. Yoder
Ronald J. Zaleski
140
MINUTES
OFTHE
1995 ANNUAL
MEETING
NEW ASSOCIATES
Karen L. Barrett
Lisa A. Bjorkman
Barry E. Blodgett
Christopher L. Bowen
Tobias E. Bradley
Michael D. Brannon
StevenA. Briggs
PamelaA. Burt
Michelle L. Busch
Martin Carrier
Victoria J. Carter
Darrel W. Chvoy
Maryellen J. Coggins
Brian C. Cornelison
Claudia M. Barry
Cunniff
Angela M. Cuonzo
Malcolm H. Curry
CharlesA.
Dal Corobbo
Dean P. Dorman
Barry P. Drobes
Mary Ann DuchnaSavrin
Jeffrey Eddinger
William P.Fisanick
Kay L. Frerk
Gary J. Ganci
Thomas P.Gibbons
StewartH. Gleason
Annette J. Goodreau
Mark A. Gorham
Monica A. Grill0
Brian D. Haney
Adam D. Hartman
Scott J. Hartzler
Daniel F. Henke
Gloria A. Huberman
David D. Hudson
Randall A. Jacobson
SuzanneG. James
Brian J. Janitschke
Philip W. Jeffery
Michael S. Johnson
Philip A. Kane IV
Ira M. Kaplan
Hsien-Ming K. Keh
Timothy P Kenetick
RobertW. Kirklin
ThereseA. Klodnicki
SalvatoreT. LaDuca
William J. Lakins
JoseeLambert
ThomasC. Lee
Isabelle Lemay
Charles R. Lenz
JamesR. Merz
Kathleen C. Odomirok
Dmitry Papush
CharlesPare
Brenda L. Reddick
Dennis L. Rivenburgh
PeterA. Royek
JasonL. Russ
ThomasA. Ryan
Manalur S. Sandilya
Michael C. Schmitz
Craig J. Scukas
Terry M. Seckel
Raleigh R. Skaggs
1~.Kevin Smith
Lori A. Snyder
ThomasStruppeck
YuanYew Tan
ThomasA. Trocchia
Jennifer S. Vincent
IsabelleT. Wang
Jeffrey D. White
David L. Whitley
Kirby W. Wisian
Mark L. Woods
Floyd %I.Yager
Mr. Kaufman then introduced Michael A. Walters, who presented the Address to New Members.
Michael J. Miller then presented the 1995 CAS Matthew
Rodermund Service Award to Dale A. Nelson. Nelson became a
MINUTESOFTHEl995ANNUALMEETING
141
Fellow of the CAS in 1965, and has volunteered more than 20
years of service to the Society throughout his career.
A moment of silence was held to mark the passing of three
members of the CAS during the past year: Kenneth L. McIntosh
(ACAS 1961), James W. Thomas (FCAS 1956), and Robert W.
Parlin (FCAS 1960).
Paul Braithwaite, CAS Vice President-Administration,
presented highlights of the Administration Report, and the Financial
Report.
Alice H. Gannon, CAS Vice President of Programs and Communications, presented the highlights of the program.
David L. Miller, chairperson of the CAS Committee on Review
of Papers, announced that one Proceedings paper would be presented at this meeting and one discussion of a previously presented Proceedings paper would be presented. In addition, three
discussions and one author’s reply would not be presented,
but published in the 1995 edition of the Proceedings, Volume
LXXXII.
The Proceedings paper and discussion presented at this meeting were:
1. “Balancing Transaction Costs and Risk Load in Risk Sharing Arrangements” by Clive L. Keatinge
2. Discussion of “A Note on the Gap Between Target and
Expected Underwriting Profit Margins” by Emilio C.
Venezian (PCAS LXXIV, November, 1987) Discussion by
William R. Gillam
The three discussions and one author’s reply not presented at
this meeting, but to be published in this volume of the Proceedings, are
1. Discussion of “Unbiased Loss Development Factors” by
Daniel F. Murphy (PCAS LXXXI, November 1994), Discussion by Daniel F. Gogol
142
MINUTES
2.
_3 .
4.
OFTHE
lYY5 AXNI
\I. hll:f:‘l I%(;
Discussion of “Risk Loads for Insurers” by Sholom Feldblum (PCAS LXXVII. November 1990), Discussion by
Todd R. Bault
Discussion “A Simulation Test of Prediction Errors of Loss
Reserve Estimation Techniques” by James Stanard (PCAS
LXXII, May 1985) Discussion by Edward Peck
Author’s Reply: “Risk Loads for Insurers” by Sholom
Feldblum
David Miller gave a brief summary of the Proceedings papers,
discussions, and author’s reply. and authors in the audience were
recognized.
David Miller then presented the 1995 CAS Dorweiler Prize to
Roger M. Hayne for his paper “Extended Service Contracts,”
which was published in the Proceedings of the Casualty Actuarial
Socieq, 1994 edition, Volume LXXXI. Miller announced that the
Woodward-Fondiller Prize would not be awarded this year.
Mr. Kaufman then concluded the business session of the Annual Meeting.
After a refreshment break, Mr. Kaufman introduced Peter Huber, who gave the Keynote Address. Huber is a lawyer, and writer,
who earned a Ph.D. from the Massachusetts Institute of Technology in mechanical engineering and a law degree from Harvard
Law School.
The first general session was held from 1I:00 a.m. to 12:30
p.m.:
The Property/Casualty Insurance Industry: A New Frontier
Moderator: Mary R. Hennessy
Panelists: J. Christopher Bulger
President
Sedgwick James of California, Inc.
MlNUTESOFTHE1995ANNUALMEETlNG
143
Bradley E. Cooper
Principal
Insurance Partners
Gary K. Ransom
Conning & Company
The general session was followed by a luncheon with the Presidential Address by Allan M. Kaufman. The luncheon was held
from 1230 p.m. to 2:00 p.m.
The afternoon was devoted to concurrent sessions, which consisted of various panels and papers.
The panel presentations covered the following topics:
1. Modeling Financial Solvency
Moderator: Oakley E. Van Slyke
President
Oakley E. Van Slyke, Inc.
Panelists: Rodney E. Kreps
Sedgwick Payne Company
William R. Van Ark
Actuary
The Wyatt Company
2.
CAS Dynamic Financial Analysis Handbook
Moderator: Susan T. Szkoda
Second Vice President and Actuary
The Travelers Insurance Company
Panelists: James K. Christie
President
IA0 Actuarial Consulting Services
Roger M. Hayne
Consulting Actuary
144
MEETING
Donald K. Rainey
Actuary
Steven T. Morgan
Vice President
3.
4.
5.
Derivatives and Reinsurance
Moderator: David Koegel
Gill & Roeser, Inc.
Panelists: Robert Arvanitis
Guy Carpenter & Company. Inc.
Sylvie Bouriaux
Senior Economist
Chicago Board of Trade
Jonathan S. Roberts
AIG Risk Management, Inc.
Update on Lloyd’s
Moderator: Paul A. Jardine
Partner
Peter K. Demmerle
Panelist:
Partner
LeBoeuf, Lamb, Greene & MacRae, LLP
Auto Insurance Fraud Weapons
Panelists: Daniel J. Johnston
President
SIINLITES OFTHE 19% ASSY-\L
6.
h1EETING
Michael L. Powell
Regional Vice President-Western Region
National Insurance Crime Bureau
Long Range Planning for the CAS
Moderator: Patrick J. Grannan
Consulting Actuary
Milliman & Robertson. Inc.
Panelists: Robert S. Miccolis
Reliance Reinsurance Corporation
Deborah M. Rosenberg
Assistant Chief Casualty Actuary
New York State Insurance Department
7.
Catastrophe Model Output in Ratemaking
Moderator: Christopher S. Carlson
Senior Actuarial Officer
Nationwide Insurance Company
Panelists: Beth E. Fitzgerald
Manager and Senior Associate Actuary
Insurance Services Oftice, Inc.
Michael A. Walters
Consulting Actuary
Tillinghast - Towers Perrin
Debra L. Werland
Executive Director
United Services Automobile Association
8.
Asset/Liability Management
Moderator: Stephen T. Morgan
Vice President
145
146
MINUTES OFTHE
Panelist:
9.
1995 ANNUAL
MEETING
Bennett W. Golub
Partner
Black Rock Financial Management
Health Care Reform
Moderator: John M. Bertko
Principal
Panelists: Louis A. Kent
Vice President and Chief Actuary
Blue Shield of California
Richard D. Schug
Actuary
10. The Cost of Reinsurance in Pricing Insurance Products
Moderator: Israel Krakowski
Actuary
Allstate Insurance Company
Panelists: Jerome E. Tuttle
Mercantile & General Reinsurance Company
Russell S. Fisher
Vice President
General Reinsurance Company
11. Environmental Liability and Superfund
Moderator: Brian Z. Brown
Consulting Actuaq
Milliman & Robertson. Inc.
Panelists:
Todd J. Hess
Senior Vice President and Chief Actuaq
Underwriters Reinsurance Compaq
Roger Carrick
Attorney
Preston, Gates & Ellis
MlNUTESOFTHE1995ANNUALMEETING
147
12. Introduction to the CAS Examination Committee
Moderator: Richard P. Yocius
Actuary
Panelists: David R. Chemick
Senior Actuary
Beth E. Fitzgerald
Manager and Senior Associate Actuary
David H. Hays
Actuary
State Farm Fire & Casualty Company
Michele A. Lombard0
Examination & Information Systems
Administrator
Virginia R. Prevosto
The officers held a reception for new Fellows and their guests
from 5:45 p.m. to 6:30 p.m. There was a general reception for all
members from 6:30 p.m. to 7:30 p.m.
Tuesday, November 14, 1995
Two general sessions were held from 8:30 a.m. to 9:30 a.m.
They were:
“NAIC and State Legislators”
Moderator: Mavis A. Walters
Executive Vice President
148
MINUTES
Panelists:
OF THE 1995 ANNUAL
MEETING
Steven T. Foster
Insurance Commissioner
Virginia Bureau of Insurance
Phillip L. Schwartz
Vice President and Associate General Counsel
American Insurance Association
Therese M. Vaughan
Insurance Commissioner
Iowa Insurance Department
The Honorable Guy Velella (R-Bronx)
Chairman/Senate Insurance Committee
State of New York
“The Cost of Capital Issues”
Moderator: Michael J. Miller
Consulting Actuary
Miller, Rapp, Herbers, Brubaker & Terry, Inc.
Richard A. Derrig
Panelists:
Steven G. Lehmann
Actuary
Company
Oakley E. Van Slyke
President
Oakley E. Van Slyke, Inc.
From 10:00 a.m. to 11:30 a.m., several concurrent sessions
were held. The panel presentations, in addition to some of the subjects covered on Monday, covered the topics of:
1. Evaluating Workers Compensation Reforms
Moderator: Robert N. Darby, Jr.
Consulting Actuary
Tillinghast - Towers Perrin
MINUTESOFTHE1995ANNUALMEETING
Panelists:
William J. Miller
Vice President
National Council on Compensation Insurance
Wade T. Overgaard
Associate Actuary
2.
California Landscape
Moderator: Richard J. Roth, Jr.
Assistant Commissioner
California Department of Insurance
Panelists: David M. Bellusci
Workers Compensation Rating Bureau of
California
John P. Drennan
3. CAS Actuarial Research Comer
Moderator: Glenn G. Meyers
4.
149
ASB Standard of Practice-Rate of Return/Profit
Provision
Moderator: Mark Whitman
Panelists: Steven G. Lehmann
Actuary
Company
Richard G. Woll
Senior Actuary
Allstate Research and Planning Center
MINL’TES OF THE 1995 AN.NU,4L. hlEETING
150
The afternoon was reserved for committee meetings and tournaments.
All members and guests enjoyed a buffet dinner, with a special
guest appearance by “Marilyn Monroe” at the “Some Like it Hot”
reception, held from 6:30 p.m. to 10:00 p.m.
Wednesday, November 15, I995
From 8:00 a.m. to 9:30 a.m., several concurrent sessions were
held and two Proceedings papers were presented. In addition to
the concurrent sessions repeated from the previous two days, the
new concurrent sessions held were:
I. Artificial Intelligence Applications in Reserving
Moderator: Roger M. Hayne
Consulting Actuary
Panelists:
i.?
Moses Cheung
President
Oxford Group
Evan Fenton
Principal
Deloitte & Touche LLP
Mark W. Mulvaney
Consulting Actuary
AAA Casualty Practice Council
Moderator: David P. Flynn
Director
First Quadrant Corporation
Panelists: Jan A. Lommele
Principal
Deloitte & Touche LLP
MlNUTESOFTHEl995ANNUALMEETtNG
IS1
Paul G. O’Connell
Director
Jean K. Rosales
Assistant Director of Government Information
American Academy of Actuaries
The Proceedings papers that were presented were:
1. “Balancing Transaction Costs and Risk Load in Risk Sharing Arrangements”
Clive L. Keatinge
Author:
2. Discussion of “A Note on the Gap Between Target and
Expected Underwriting Profit Margins,” PCAS LXXIV,
1987, by Emilio C. Venezian
William R. Gillam
Author:
Following the concurrent sessions, William W. Palmer, General
Counsel, California Department of Insurance, gave a special presentation to the CAS members.
From 10:00 a.m. to 11:30 a.m., the final general session was
held.
“Whither Liability Reform?’
Moderator: Michael L. Toothman
Partner
Arthur Andersen LLP
Panelists: Jeffrey Gifford
Pavalon & Gifford, PC.
Sherman Joyce
President
American Tort Reform Association
Philip D. Miller
Consulting Actuary
Tillinghast -Towers Perrin
152
M[NUTES OF THE 1995 ANNUAL
MEETING
After the official passing of the presidential gavel from outgoing CAS President Allan M. Kaufman to new CAS President
Albert J. Beer, Mr. Kaufman announced future CAS meetings,
gave closing remarks, and officially adjourned the 1995 CAS
Annual Meeting at 11:45 a.m.
November
I995 Attendees
The 1995 CAS Annual Meeting was attended by 455 Fellows.
242 Associates, and 293 Guests. The names of the Fellows and
Associates in attendance follow:
FELLOWS
BarbaraJ. Addie
Rhonda K. Aikens
Kristen M. Albright
Terry J. Alfuth
Jean-Luc E. Allard
Kerry F. Allison
Richard R. Anderson
Charles M. Angel1
Robert A. Anker
Kenneth Apfel
John G. Aquino
Richard V. Atkinson
William M. Atkinson
Guy A. Avagliano
Karen F. Ayres
Anthony J. Balchunas
Timothy J. Banick
W. Brian Barnes
Todd R. Bault
Edward J. Baum
Philip A. Baum
Gregory S. Beaulieu
Allan R. Becker
Albert J. Beer
Linda L. Bell
David M. Bellusci
William H. Belvin
Phillip N. Ben-Zvi
ReginaM. Berens
JamesR. Berquist
G. Gregory Bertles
Neil A. Bethel
Richard A. Bill
JamesE. Biller
Richard S. Biondi
Everett G. Bishop
Michael P. Blivess
Gary Blumsohn
LeRoy A. Boison, Jr.
Paul Boisvert, Jr.
Ronald L. Bornhuetter
Wallis A. Boyd, Jr.
Nancy A. Braithwaite
Paul Braithwaite
JamesF. Brannigan
Donna D. Brasley
YaakovB. Brauner
Paul J. Brehm
Dale L. Brooks
J. Eric Brosius
Brian Z. Brown
William W. Brown, Jr.
Randall E. Brubaker
Gary S. Bujaucius
Mark E. Burgess
Patrick J. Burns
Mark W. Callahan
ClaudetteCantin
Ruy A. Cardoso
Christopher S. Carlson
Lynn R. Carroll
Michael J. Cascio
Martin Cauchon
Michael J. Caulfield
Kevin J. Cawley
Ralph M. Cellars
Galina M. Center
Francis D. Cerasoli
JanetL. Chaffee
Scott K. Charbonneau
David R. Chernick
JamesK. Christie
Allan Chuck
Gregory J. Ciezadlo
Laura R. Claude
Michael A. Coca
Robert F. Conger
EugeneC. Connell
Mary L. Corbett
Francis X. Corr
Alan M. Crowe
Alan C. Curry
Michael T. Curtis
Daniel J. Czabaj
Ronald A. Dahlquist
David J. Darby
Robert N. Darby, Jr.
JeromeA. Degerness
Daniel Demers
Marie-Julie Demers
Patrick K. Devlin
Mark A. Doepke
Michael C. Dolan
John P. Doucette
John P. Drennan
Diane Symnoski Duda
Richard Q. Easton
Maribeth Ebert
Dale R. Edlefson
Bob D. Effinger, Jr.
Gary J. Egnasko
Valere M. Egnasko
MEETING
Nancy R. Einck
ThomasJ. Ellefson
JohnW. Ellingrod
JamesEly
JeffreyA. Englander
David Engles
Paul E. Ericksen
Dianne L. Estrada
Glenn A. Evans
Philip A. Evensen
John S. Ewert
Doreen S. Faga
JanetL. Fagan
Madelyn C. Faggella
Michael A. Falcone
Dennis D. Fasking
DeniseA. Feder
Mark E. Fiebrink
Russell S. Fisher
Beth E. Fitzgerald
Daniel J. Flaherty
David P Flynn
Claudia S. Forde
David C. Forker
Barry A. Franklin
Kenneth R. Frohlich
Michael Fusco
Alice H. Gannon
Andrea Gardner
Louis Gariepy
JamesJ. Gebhard
David B. Gelinne
John F. Gibson
Richard N. Gibson
Bonnie S. Gill
William R. Gillam
153
Mary K. Gise
Bradley J. Gleason
Ronald E. Glenn
StevenA. Glicksman
Daniel C. Goddard
StevenF. Goldberg
CharlesT. Goldie
Leon R. Gottlieb
SusanM. Gozzo
Andrews
Gregory S. Grace
David J. Grady
Marc C. Grandisson
Patrick J. Grannan
Anne G. Greenwalt
StevenJ. Groeschen
Linda M. Groh
Carleton R. Grose
Marshall J. Grossack
Denis G. Guenthner
David N. Hafling
JamesW. Haidu
Allen A. Hall
JamesA. Hall IlI
Robert C. Hallstrom
Malcolm R. Handte
William D. Hansen
H. Donald Hanson
JonathanM. Harbus
Christopher L. Harris
David C. Harrison
David G. Hartman
Matthew T. Hayden
Roger M. Hayne
David H. Hays
154
SuzanneE. Henderson
Mary R. Hennessy
TeresaJ. Herderick
Todd J. Hess
David R. Heyman
Carlton W. Honebein
Deborah G. Horovitz
Paul E. Hough
GeorgeA. Hroziencik
Anthony Iafrate
Patrick C. Jensen
Russell T. John
Larry D. Johnson
Thomas S. Johnston
Jeffrey R. Jordan
Adrienne B. Kane
Frank J. Karlinski 111
Janet S. Katz
Allan M. Kaufman
Clive L. Keatinge
Eric R. Keen
Tony J. Kellner
Anne E. Kelly
Brian Danforth Kemp
Deborah E. Kenyon
Allan A. Kerin
Frederick W.
Kilbourne
Joe C. Kim
Michael F. Klein
Joel M. Kleinman
Charles D. Kline, Jr.
Douglas F. Kline
Paul J. Kneuer
Terry A. Knull
Leon W. Koch
John JosephKollar
Mikhael I. Koski
ThomasJ. Kozik
Israel Krakowski
GustaveA. Krause
Rodney E. Kreps
David J. Kretsch
Adam J. Kreuser
JaneJasperKrumrie
John R. Kryczka
Jeffrey L. Kucera
Andrew E. Kudera
David R. Kunze
Michael A. LaMonica
Blair W. Laddusau
D. Scott Lamb
DeanK. Lamb
JohnA. Lamb
Alan E. Lange
JamesW. Larkin
Michael D. Larson
PaulW. Lavrey
Paul B. LeStourgeon
Nicholas M.
Leccese,Jr.
Robert H. Lee
Marc-Andre Lefebvre
StevenG. Lehmann
JosephW. Levin
Aaron S. Levine
GeorgeM. Levine
Allen Leu
John J. Leivandouski
StephanieJ. Lippl
Barry C. Lipton
Richard W. Lo
JanA. Lommele
Edward P. Lotkowski
W. JamesMacGinnitie
Maria Mahon
BarbaraS. Mahoney
Lawrence F. Marcus
Joseph0. Marker
StevenD. Marks
Blaine C. Marles
StevenE. Math
RobertW. Matthews
Michael G. McCarter
Robert D. McCarthy
John W.
McCutcheon,Jr.
William G. McGovern
Michael F. McManus
Michael A. McMurray
Dennis T. McNeese
Dennis C. Mealy
William T. Mech
John P.Mentz
Robert E. Meyer
StephenJ. Meyer
Glenn G. Meyers
Robert S. Miccolis
David L. Miller
David L. Miller
Michael J. Miller
Philip D. Miller
Ronald R. hliller
William 3. Miller
Neil B. Miner
CharlesB. Mitzel
MlNUTESOFTHEl995ANNUALMEETlNG
Frederic JamesMohl
David F. Mohrman
Bruce D. Moore
Kelly L. Moore
Russell E. Moore
Francois Morin
Jay B. Morrow
Robert V. Mucci
Evelyn Toni Mulder
Mark W. Mulvaney
Richard E. Munro
John A. Murad
Daniel M. Murphy
William F. Murphy
Thomas E. Murrin
Antoine A. Neghaiwi
JamesR. Neidermyer
Chris E. Nelson
Dale A. Nelson
Kenneth 1. Nemlick
Karen L. Nester
Richard T. Newell, Jr.
John Nissenbaum
Ray E. Niswander,Jr.
Victor A. Njakou
Kathleen C. Nomicos
StephenR. Noonan
Keith R. Nystrom
Kathy A. Olcese
Bruce E. Ollodart
Paul M. Otteson
JoanneM. Ottone
WadeT. Overgaard
Timothy A. Paddock
Richard D. Pagnozzi
Gary S. Patrik
SusanJ. Patschak
WendeA. Pemrick
Melanie Turvill
Pennington
CharlesI. Petit
Mark W. Phillips
JosephW. Pitts
Arthur C. Placek
Richard C. Plunkett
Denis Poirier
On CheongPoon
Jeffrey H. Post
JosephJ. Pratt
Philip 0. Presley
Virginia R. Prevosto
Mark Priven
Boris Privman
Arlie J. Proctor
Glenn J. Pruiksma
John M. Purple
Mark S. Quigley
Richard A. Quintano
Kay K. Rahardjo
Donald K. Rainey
RajagopalanK. Raman
Gary K. Ransom
Jerry W. Rapp
Donna J. Reed
JamesF. Richardson
Donald A. Riggins
Richard D. Robinson
StevenCarl Rominske
Allen D. Rosenbach
DeborahM. Rosenberg
Kevin D. Rosenstein
Gail M. Ross
155
Richard J. Roth, Jr.
Bradley H. Rowe
John M. Ruane,Jr.
JamesV. Russell
Stuart G. Sadwin
ThomasE. Schadler
Harold N. Schneider
David C. Scholl
Richard D. Schug
RogerA. Schultz
JosephR. Schumi
Kim A. Scott
PeterSenak
Vincent M. Senia
Derrick D. Shannon
Alan R. Sheppard
Harvey A. Sherman
Richard E. Sherman
Edward C. Shoop
Melvin S. Silver
Christy L. Simon
Rial R. Simons
David Skurnick
John Slusarski
Christopher M.
Smerald
Lee M. Smith
Richard A. Smith
Richard H. Snader
David B. Sommer
Keith R. Spalding
JoanneS. Spalla
Bruce R. Spidell
Daniel L. Splitt
Douglas W. Stang
Elton A. Stephenson
156
Richard A. Stock
Edward C. Stone
JamesP. Streff
Douglas N. Strommen
Stuart B. Suchoff
Mary T. Sullivan
JamesSurrago
Russel L. Sutter
John A. Swift
SusanT. Szkoda
Catherine Harwood
Taylor
Frank C. Taylor
Kathleen W. Terrill
Karen F. Terry
Richard D. Thomas
Kevin B. Thompson
Ernest S. Tistan
Thomas C. Tote
Darlene P Tom
Charles F. Toney II
Michael L. Toothman
Cynthia J. Traczyk
Nancy R. Treitel
Frank J. Tresco
Michel Trudeau
Everett J. Truttmann
WarrenB. Tucker
Patrick N. Tures
JeromeE. Tuttle
PeterS. Valentine
William R. VanArk
Charles E.
Van Kampen
David B. Van
Koevering
Kenneth R.
Van Laar, Jr.
Oakley E. Van Slyke
Mark D. van Zanden
Trent R. Vaughn
Ricardo Verges
Gerald R. Visintine
JosephL. Volponi
William J. VonSeggern
Christopher P.Walker
Glenn M. Walker
ThomasA. Wallace
Lisa Marie Walsh
Mavis A. Walters
Michael A. Walters
Kelly A. Wargo
ThomasV. WarthenIII
Nina H. Webb
Dominic A. Weber
Patricia J. Webster
ThomasA. Weidman
John l? Welch
L. Nicholas
Weltmann,Jr.
Debra L. Werland
Robert G. Whitlock, Jr.
Mark Whitman
Gregory S. Wilson
Chad C. Wischmeyer
Michael L. Wiseman
Richard G. Woll
Patrick B. Woods
John S. Wright
Walter C. Wright III
Paul E. Wulterkens
Richard P.Yocius
ClaudeD. Yoder
HeatherE. Yow
JamesW. Yow
Ronald J. Zaleski
ASSOCIATES
Mark A. Addiego
JonathanD. Adkisson
Elise M. Ahearn
John P.Alltop
Larry D. Anderson
JamesA. Andler
Michael E. Angelina
William P.Ayres
Karen L. Barrett
Douglas S. Benedict
Cynthia A. Bentley
Bruce J. Begeron
Lisa A. Bjorkman
Barry E. Blodgett
Thomas S. Boardman
GeorgeP.Bradley
Tobias E. Bradley
Michael D. Brannon
StevenA. Briggs
PeterV. Burchett
MINUTESOFTHE1995ANNUALMEETlNG
William E. Burns
Jeff Eddinger
PamelaA. Burt
David M. Elkins
Richard F. Burt, Jr.
CharlesV. Faerber
Michelle L. Busch
Kendra M. FeliskyTara E. Bush
Watson
Kenrick A. Campbell
Bruce D. Fell
Victoria J. Carter
Carole M. Ferrer0
Julia C. Causbie
David N. Fields
Darrel W. Chvoy
William P.Fisanick
Gary T. Ciardiello
Robert E Flannery
Kay A. Cleary
Daniel J. Flick
J. Paul Cochran
Kay L. Frerk
Jo Ellen Cockley
Kai Y. Fung
Mary B. Gaillard
Howard S. Cohen
Gary J. Ganci
ThomasP.Gibbons
Karl D. Colgren
Michael A. Ginnelly
Vincent P.Connor
Nicholas P Giuntini
Brian C. Cornelison
Kirsten J. Costello
StewartH. Gleason
ChristopherG. Cunniff StevenB. Goldberg
Terry L. Goldberg
Claudia M. Barry
Cunniff
Annette J. Goodreau
Mark A. Gorham
Angela M. Cuonzo
Gary Granoff
Malcolm H. Curry
CharlesA.
Bruce H. Green
Dal Corobbo
Monica A. Grill0
ThomasV. Daley
William Alan Guffey
Michael K. Daly
Michele l? Gust
Jeffrey F. Deigl
Leigh JosephHalliwell
William Der
Aaron Halpert
Gordon F. Diss
PaulJamesHancock
Frank H. Douglas
Brian D. Haney
RobertL.
William A. Dowel1
Kimberly J. Drennan
Harnatkiewicz
Adam D. Hartman
Barry P.Drobes
BernardDuPont
Scott J. Hartzler
157
Barton W. Hedges
Daniel F. Henke
JosephP Henkes
Paul D. Henning
JosephA. Herbers
Bernard R. Horovitz
Gloria A. Huberman
David D. Hudson
Jeffrey R. Hughes
Jeffrey R. Ill
Randall A. Jacobson
Brian J. Janitschke
Fong-YeeJ. Jao
Philip W. Jeffery
Daniel J. Johnston
JamesW. Jonske
Edwin G. Jordan
Philip A. Kane IV
Ira M. Kaplan
CharlesN. Kasmer
David L. Kaufman
Hsien-Ming K. Keh
StevenA. Kelner
Timothy P Kenefick
RebeccaA. Kennedy
SusanE. Kent
Ann L. Kiefer
Robert W. Kirklin
ThereseA. Klodnicki
David Koegel
Louis K. Korth
KennethAllen
Kurtzman
SalvatoreT. LaDuca
David W. Lacefield
William J. Lakins
IS8
SIINLITES OFTHE
JoseeLambert
Matthew G. Lange
Thomas C. Lee
Elizabeth Ann
Lemaster
Isabelle Lemay
Charles R. Lenz
Sam F. Licitra
Richard B. Lord
David J. Macesic
SudershanMalik
Donald E. Manis
Anthony L. Manzitto
Gabriel 0. Maravankin
Janice L. Markc
Leslie R. Marlo
Jeffrey F. McCarty
StephenJ. McGee
EugeneMcGovern
Donald R. McKay
StephenV. Merkey
JamesR. Mere
Linda K. Miller
Neil L. Millman
Paul W. Mills
Stanley K. Miyao
StephenThomas
Morgan
Donald R. Musante
Mark Naigles
John K. Nelson
Aaron WestNewhoff
Henry E. Newman
PeterM. Nonken
Dale F. Ogden
I’)‘)5 .\VV141.
hll:IKl I%(;
DouglasW. Oliver
Richard A. Olsen
Charles P.Orlowicz
TeresaK. Paffenback
Dmitry E. Papush
Charles Pare
ThomasPassante
Clifford A. Pence,Jr.
Richard A. Plano
Katherine D. Porter
Ruth Youngner
Poutanen
Michael D. Price
Regina M. Puglisi
Ralph StephenPuli<
Karen L. Queen
Kathleen Mary Quinn
Eric K. Rabenold
Yves Raymond
JamesE. Rech
BrendaL. Reddick
StevenJ. Regnier
Ellen J. Respler
Victor Unson Revilla
Dennis L.
Rivenburgh,Jr.
Douglas S. Rivenburgh
JonathanS. Roberts
Scott J. Roth
JamesB. Rowland
PeterA. Royek
Michael R. Rozema
GeorgeA. Rudduck
JasonL. Russ
David A. Russell
StephenP.Russell
ThomasA. Ryan
SandraSamson
,ManalurS. Sandilya
slichael Sansevero,Jr.
SandraC. Santomenno
StephenPaul Sauthoff
Michael C. Schmitz
Frederic F. Schnapp
PeterR. Schwanke
Craig J. Scukas
Terry M. Seckel
Ahmad ShadmanValuvi
TheodoreR. Shalack
Kerry S. Shubat
JanetK. Silverman
Jeffrey S. Sirkin
Raleigh R. Skaggs,Jr.
Byron W. Smith
Gina L.B. Smith
L. Kevin Smith
David C. Snou
Lori A. Snyder
John A. Stenmark
Michael J. StewardII
llene G. Stone
Thomas Struppeck
Joy Y. Takahashi
Craig P.Taylor
Joseph0. Thorne
John P.Thorrick
ThomasA. Trocchia
Robert C. Turner, Jr.
JamesF. Tygh
Frederick A. Urschel
ThereseVaughan
MINUTES OFTHE
Jennifer S. Vincent
JeromeF. Vogel
JosephW. Wallen
Monty J. Washburn
Russell B. Wenitsky
1995 ANNUAL
MEETING
Jeffrey D. White
David L. Whitley
Elizabeth R. Wiesner
Kirby W. Wisian
Robert I? Wolf
Cheng-ShengP.Wu
Floyd M. Yager
Robert S. Yenke
Vincent F. Yezzi
Barry C. Zurbuchen
159
REPORT OF THE VICE PRESIDENT-ADMINISTRATION
The objective of this report is to provide a brief summary of
CAS activities since the last CAS Annual Meeting.
I will first comment on these activities as they relate to the
following purposes of the Casualty Actuarial Society as stated in
our Constitution:
1. Advance the body of knowledge of actuarial science in
applications other than life insurance:
2. Establish and maintain
membership;
standards of qualification
for
3. Promote and maintain high standards of conduct and
competence for the members; and
4. Increase the awareness of actuarial science.
I will then provide a summary of other activities that may not
relate to a specific purpose but yet are critical to the ongoing
vitality of the CAS. And lastly, I will update you on the current
status of our finances and key membership statistics.
In support of Purpose 1, the CAS has devoted significant resources during the past year to initiating research, with specific
focus on developing the concept of Dynamic Financial Analysis
(DFA). As one indication of the increased pace of activity this
year, the CAS currently has eight different prize programs and
funded research projects in various stages of progress.
The high priority areas of research during the year included
the following projects as assigned to the appropriate committee:
l
l
A funded research project to identify variables used in financial analysis models (DFA Task Force on Variables),
Publication
Financial
in September 1995 of the Handhook OIEDynamic
Analysis
for
Property/Casualty
Insurance
(Committee on Valuation and Financial Analysis),
160
Companies
REPORT OF TlIE VICE PRESIDENT-ADMINISTRATlON
l
l
l
161
Completion of a draft Statement of Guidance Regarding Management Data and Information (Committee on Management
Data and Information).
Development of a 1996 Call Paper Program on Ratemaking
(Committee on Ratemaking).
Drafting of a proposed exposure draft on Principles of Risk
Classification (Committee on Risk Classification).
New papers published in the Proceedings. the Forum, and
the other CAS publications all increase the body of knowledge
available to our profession. An issue of the Forum released in
Fall 1995 included research reports by both CAS and American Academy of Actuaries committees as well as a number of
papers from actuaries and academics. The 1994 Proceedings included seven new papers on a variety of topics.
Continuing education opportunities help fulfill Purpose 3, and
a significant amount of DFA material was offered in this year’s
programs. Seminars relating to DFA topics included the first special interest seminar devoted specifically to DFA, held in July
1995; the limited attendance seminar on “Principles of Finance
in Pricing Property and Casualty Insurance” which was held
twice; a new limited attendance seminar on “Managing Asset
and Investment Risk;” and the CIA/GAS Seminar for the Appointed Actuary held in September 1995. The CAS Spring Meeting in May featured eight papers presented as part of the discussion paper program on “Incorporating Risk Factors in Dynamic
Financial Analysis.” In addition to this meeting, the Casualty Loss Reserve Seminar, the June Reinsurance Seminar, the
Ratemaking Seminar, and the CAS Annual Meeting all contained sessions on DFA subjects.
The Admissions Committees provide the major support for
Purpose 2. They make continuous improvements to the syllabus
and exam preparation and grading process, while overseeing the
administration of the testing of approximately 6,500 registered
162
REPORT 01; THE VICE PRESIDfiN’I-At~MINIS’~l~~I’ION
candidates. Major initiatives in this area this year included:
Establishment of an Educational Task Force to identify needed
skills for actuaries of the future.
l
Adoption of separate Canadian and U.S. revisions of Part 7.
These separate exams will include three categories of syllabus
material: common core material, common material with different testing emphasis, and nation-specific material. Part 8 has
been restructured in the same categories.
l
Introduction
of a Student Liaison Committee to improve
communications between students and admissions committees. In order to facilitate communications, a quarterly Student Newsletter is under development, with an inaugural issue
planned for distribution in December 1995.
l
A report by the Travel Time Working Group was completed
and presented to the Board of Directors in February 1995. This
report. recently published in the CAS Forum. represents the
culmination of a two-year effort to establish the information
necessary for the CAS to monitor travel time through the
CAS exam process; ensure that the CAS database contains the
define the criteria by which travel
requisite information;
time should be monitored; and draw preliminary conclusions
regarding the impact of exam partitions on travel time. if possible.
Maintaining our high standards is also accomplished through
a quality program of continuing education. The CAS provides
these opportunities through the publication of actuarial materials
and the sponsorship of a number of meetings and seminars. This
year’s sessions included:
l
l
The Spring and Annual Meetings, held in St. Iouis and San
Diego;
The 1995 CAS Seminar on Ratemaking. held in New Orleans,
which had 733 registrants;
163
l
l
l
l
l
l
l
The Casualty Loss Reserve Seminar in Chicago, of which the
CAS is a co-sponsor with the American Academy of Actuaries, attended by 753;
The special interest seminar on “Profitability”
tended by 96;
in April
at-
The special interest seminar on “Dynamic Financial Analysis”
in July in Atlanta, attended by 135;
The special interest seminar on “Environmental
October in Boston, attended by 136;
Risks” held in
The Reinsurance Seminar in June with attendance of 328 in
New York City;
The CIA/GAS Seminar for the Appointed Actuary in Montreal, co-sponsored by the Canadian Institute of Actuaries and
the CAS, attended by 290; and
The previously
DFA topics.
mentioned
limited
attendance seminars on
The Continuing Education Committees have explored ways to
provide additional opportunities to our membership. This year,
we saw continued growth in the use of a relatively new forum:
limited attendance seminars with academic instructors. These
have been well received.
The CAS Regional Affiliates also provide valuable opportunities for the members to participate in educational forums. In
addition, the Regional Affiliates are a resource to help increase
the awareness of the profession (Purpose 4) at the local level.
Discussions are underway with the leadership of the Regional
Affiliates to encourage more communication at the high school
level.
The CAS also promoted awareness of the profession through
continued financial support of the Forecast 2000 program. This
program seeks to align the actuarial profession with crucial pub-
164
KEPOKr OF THE VICE PRESIIXNl-AI~MINIS’IRATION
lit policy issues and increase visibility
eral public.
of actuaries with the gen-
Also related to the fourth purpose, but generally impacting all purposes, are the CAS’s international activities. In addition to the ongoing attendance at various international actuarial
society meetings by the CAS leadership, the CAS became a
charter member of the International Forum of Actuarial Associations.
The CAS Office continues to provide excellent support and to
expand its services and capabilities. Significant productivity gains
have been realized with their enhanced MIS capabilities, while
support for exam administration and the annual budget process
have been greatly enhanced. New member services introduced
this year include the development of a quarterly continuing education calendar and enhancement of CAS bulletin board capabilities. The CAS Office is also providing an increased level of
support to the various committees.
Another resource of the CAS, and an integral part of its fabric and success, is its committees and many volunteers. Member
participation on our committees remains high. The annual Committee Chairpersons’ Meeting in March was highlighted by group
discussions of key CAS issues.
In closing, I will provide a brief status of our membership and
financial condition. Our size continued its rapid increase as we
added 200 new Associates and 114 new Fellows. Our membership now stands at 2,490.
New members elected to the Board of Directors for next year
include Regina M. Berens, Claudette Cantin, David R. Chernick, C. K. “Stan” Khury, and David L. Miller. The membership elected Robert A. Anker to the position of President-Elect,
while Albert J. Beer will assume the Presidency.
The Executive Council, with primary responsibility for day-today operations, met either by teleconference or in person at least
REPORT OF ME VICE PRESIDENT-ADMINKI’RATlON
165
once a month during the year. The Board of Directors elected
the following Vice Presidents for the coming year.
Vice President-Administration,
Paul Braithwaite
Vice President-Admissions, John J. Kollar
Vice President-Continuing
Education, Susan T. Szkoda
Vice President-Programs and Communications,
Patrick J. Grannan
Vice President-Research and Development, Michael J. Miller
The CPA firm of Feddeman & Company has been engaged to
examine the CAS books for fiscal year 1995 and its findings will
be reported by the Audit Committee to the Board of Directors in
February 1996. The fiscal year ended with unaudited net income
of $369,684, which compares favorably to a budgeted amount of
$38,506. Members’ equity now stands at $1,617,288, subdivided
as follows:
$91,292 (
Michelbacher Fund
r-- ~
Dorweiler
Fund
I
5,115 1
CAS Trust
3,469
Scholarship Fund
7,319
Rodermund Fund
13,934
CLRS Fund
5,000
ASTIN Fund
4,000
180,665
Research Fund
1,306,495
CAS Surplus
TOTAL
MEMBERS’
EQUITY
$1,617,288
This represents an increase in equity of $376,620 over the
amount reported last year.
166
l<fIf’Oft’f‘Of~‘~flfi
WC-f: Pflf:SfDf:N
I ~.zI)MINISI~</WION
For 1995-96, the Board of Directors has approved a budget of
approximately $3.2 million. Members’ dues for next year will be
$260; an increase of $10, while fees for the Invitational Program
will increase by $15 to $320.
Respectfully submitted.
Paul Braithwaite
Vice President-Adminisf rution
November 13, 1995
167
FINANCIAL
REPORT
FISCAL YEAR ENDED 9/30/95
OPERATING RESULTS BY FUNCTION
FUNCTION
INCOME
Membership Serv~cs
Semmars
Meetings
Exnmq
Publications
TOTA L
NOTES:
EXPENSE
DIFFERENCE
$
704.013
S 132,492
$ (28.479)
69 I.780
435.147
2.56.633
481.735
435,788
45.947
2.067.X 12
I .926,995 (n)
140.817
42.345
29,807
12,538
$ 3.987.685
$ 3.560.229
$ 427,456 (h)
(a) Includes $I ,279.OOOof volunteer serwces for income and expense.
(h) Change in surplus before interfund transfers of $S2,ooO for recewch and ASTIN funds
BALANCE SHEET
ASSE7S
Checkmg Account
T-B~llslNoter
Accrued lntewt
CLRS Deporlt
Prepaid Expenses
Prepaid Insurance
Accounlr Reccwnhle
Comouters. Fumlture
Less: Accumulated Depreciation
TOTA I. ASSETS
9/30/94
366.425
I ,197.008
19.185
S.000
5 1.694
6,628
45,000
233.279
(149.899)
$ 1,774,320
$
LIABILITIES
Exam Fees Deferred
$
Annual Meeung Fees Deferred
Semmar Fees Deferred
Llmltcd Attendance Workshop Fees Deferred
AccounL? Poyahle and Accrued Expenses
Deferred Rent
Accrued Pemion
TOTAL LIABILITIES
S
MEMBERS’
$
$
9/30/95
315.087
38.359
47,328
700
134.589
39.002
36.101
611.166
DIFFERENCE
$ (316.165)
783,036
35,476
0
(27,884)
I.321
0
26.52 I
(42,400)
$ 459.905
DIFFERENCE
s
I R.098
s
(40.38I)
16,474
700
76,256
(6.072)
12,440
77514
EQUITY
Unrertricted
CAS Surolus
CLRS F&d
Michclbacher Fund
Dorweiler Fund
CAS Trust
Research Fund
ASTIN Fund
Suhtotnl Unrestricted
Temporarily Restncted
Scholarship Fund
Rcdermund Fund
Subtotal Restricted
TOTAL
9/30/94
296.989
78.740
30,854
0
58,333
45,074
23,661
533.653
9/3o/Y5
50,260
I ,980,044
54.661
5.000
23.810
7,949
4s.KM
259.800
( 192,299)
S 2.234.225
$
EQUITY
9/30/94
936.810
5,000
87,896
5.823
3,305
178.16.5
2,000
1.218,999
9/3o/?x
$ 1.312.266
91.292
5.1 is
3,469
180.665
4.0ot7
I ,60 I.807
382.808
7,446
14.222
21.668
% 1.240.667
7.318
13.934
2 I-2.52
$ l.623.059
(128)
(288)
(416)
382.391
$
DIFFERENCE
$
5.ooo
Paul Braithwaite, Vice President-Administration
This is to certij) that the assefs and accounts show in the above
financial statement kar’e been audited and found to he correct,
CAS Audit Committee: Steven E Goldberg, Chairperson;
Robert F. Conger, Anthony J. Grippa, and William M. Rowland.
375.456
0
3.396
(70X)
I64
2.500
2mo
S
1995 EXAMINATIONS-SUCCESSFUL
CANDIDATES
Examinations for Parts 3B, 4A, 4B, 5A, 5B, 6, 8, 8C (Canadian), and 10 of the Casualty Actuarial Society were held on May
1, 2, 3, 4, and 5, 1995. Examinations for Parts 3B, 4A, 4B, SA,
5B, 7, and 9 of the Casualty Actuarial Society were held on November 1, 2, 3, and 6, 1995.
Examinations for Parts 1, 2, 3A, and 3C (SOA courses 100,
110, 120, and 135) are jointly sponsored by the Casualty Actuarial
Society and the Society of Actuaries. Parts 1 and 2 were given in
February, May, and November of 1995, and Parts 3A and 3C were
given in May and November of 1995. Candidates who were successful on these examinations were listed in joint releases of the
two societies.
The Casualty Actuarial Society and the Society of Actuaries
jointly awarded prizes to the undergraduates ranking the highest
on the Part 1 examination.
For the February 1995 Part 1 examination, the $200 first prize
winners were Robert J. Aguirre of Rice University and Patrick
Beaudoin of the University of Laval. The $100 second prize winners were Sumit K. Daftuar of Harvard University, Erik J.
Sandquist of Cornell University, and Jue Wang of Queens College.
For the May 1995 Part 1 examination, the $200 first prize was
awarded to Emil B. Kraft of Eastern Washington University. The
$100 second prize winners were Fai T. Tong of the University of
Chicago; and Hai Lin, Liqiang Ni, Hairong Zhang, and Zhisheng
Zhou, all from Fudan University in Shanghai.
For the November 1995 Part 1 examination, the $200 first prize
was awarded to Kohji Hirabayashi from the University of Tokyo
and Lifeng Wu of Fudan University in Shanghai. The $100 second
prize winners were Nicholas Albicelli, State University of New
York; Paul Colucci, University of Illinois; Daoyong Lou, Fudan
University in Shanghai; and Timothy Mosler, Florida Atlantic
University.
168
CANDIDATES
169
The following candidates were admitted as Fellows and Associates at the CAS Spring Meeting in May 1995 as a result of their
successful completion of the Society requirements in the November 1994 examinations.
NEW FELLOWS
Timothy J. Cremin
Bradley A. Granger
Mathieu Lamy
SuzanneMartin
Brett E. Miller
Mark Priven
Eduard J. Pulkstenis
John F. Rathgeber
David M. Savage
Jeffery J. Scott
Russell Steingiser
Eileen M. Sweeney
Yuan-YuanTang
ThomasC. Tote
John V. Van de Water
PeterG. Wick
NEW ASSOCIATES
Rimma Abian
John I? Alltop
K. Athula P.Alwis
StevenD. Armstrong
Martin S. Arnold
Bruce J. Bergeron
StevenL. Berman
Corey J. Bilot
Carol A. Blomstrom
John T. Bonsignore
Douglas J. Bradac
Kevin M. Brady
Betsy A. Branagan
JamesL. Bresnahan
Lisa J. Brubaker
Elliot R. Burn
Tara E. Bush
J’ne E. Byckovski
SandraL. Cagley
PamelaJ. Cagney
DouglasA. Carlone
Jill C. Cecchini
HeatherL. Chalfant
Jean-Franqois
Chalifoux
PeggyCheng
Gary C. K. Cheung
ChristopherJ. Claus
William F. Costa
ChristopherG. Cunniff
SeanR. Devlin
BehramM. Dinshaw
William A. Dowel1
Kimberly J. Drennan
Pierre Drolet
StephenC. Dugan
Tammy L. Dye
S. Anders Ericson
JamesG. Evans
Daniel J. Flick
Andre F. Fontaine
SusanT. Garnier
Christopher H. Geering
Eric J. Gesick
John T. Gleba
John E. Green
StevenA. Green
CharlesR. Grilliot
Julie K. Halper
David S. Harris
Betty-Jo Hill
John V. Hinton
JasonN. Hoffman
John E Huddleston
Li Hwan Hwang
Brian L. Ingle
Christian Jobidon
Daniel K. Johnson
Gail E. Kappeler
Lowell J. Keith
Thomas P. Kenia
170
1995 EXAMINATIONS-SUCCESSFL’L
Michael B. Kessler
Jean-Raymond
Kingsley
Gary R. Kratzen
Brian S. Krick
Marc LaPalme
Debra K. Larcher
Gregory D. Larcher
Daniel E. Lents
Edward A. Lindsay
Richard B. Lord
Cornwell H. Mah
Anthony L. Manzitto
Scott A. Martin
Tracey L. Matthew
Camley A. Mazloom
Deborah I,. Mc<‘raq
Kelly S. McKeethan
Lynne S. McWithey
Claus S. Metlner
CANDIDATES
Paul W. Mills
Anne H. Moore
Kenneth B. Morgan. Jr.
Kevin T. Murphy
Hiep T. Nguyen
JamesL. Nutting
Milary N. Olson
Thomas Passante
Nicholas H. Pastor
Claude PenlandIV
William Peter
GenevievePineau
Robert E. Quane111
PeterS. Rauner
Natalie J. Rekittke
Scott Reynolds
Meredith Cr.
Richardson
John W. Rollins
RajeshV.
Sahasrabuddhe
Marilyn E. Schafer
Michael J. Scholl
M. Kate Smith
John B. Sopkowicz
Michael J. Sperduto
Scott D. Spurgat
Scott T. Stelljes
Kevin D. Strous
StevenJ. Symon
Joy Y. Takahashi
David M. Terne
Daniel A. Tess
SonT. Tu
Eric Vaith
Cynthia L. Vidal
Rohert J. Walling III
StevenB. White
l-Ii/aheth R. Wiesner
Jlichael J. Williams
The following is a lict of succe<<ful canditlatcs in CAS examinations held in May 1995.
Part 3B
Jason R. Abrams
William J. Albertson
Anthony L. Alfieri
Genevieve L. AllenState
Frank J. Barnes
Wendy A. Barone
SuzanneBarry
Andre Beaulieu
Jennifer L. Beck
HeatherL. Bennett
Shelley L. Bitner
Lisa A. Bjorkman
JonathanE. Blake
Mary Denise
Boarman
Mark E. Bohrer
JamesG. Brad}
ChristopherS.
Bramstedt
RodneyL. Brunk
Paul E. Budde
Lisa K. Buege
Lori L. Burton
.Llichelle L. Busch
Matthew E. Butler
Kelli R. Culdwell
1995 EXAMINATIONS-SUCCESSFULCANDIDATES
Todd D. Cheema
ThomasJ. Chisholm
Lisa A. Chodaczek
CatherineChoi
Andrew K. Chu
CharlesA. Cicci
Lori Anne Cieri
Edward W. Clark
Eric J. Clymer
CostasA. Constantinou
Ellen B. Cooper
Brian C. Cornelison
Crystal Dawn Danner
Elisa J. Davenport
Barry P.Drobes
Louis Durocher
Elizabeth B. Emory
William H. Erdman
Jui-Chuan Fan
William M. Finn
Noelle C. Fries
John E. Gaines
Sherri L. Galles
Daniel J. Gieske
Moshe D. Goldberg
John P.Gots
Melanie T. Green
Daniel C. Greer
Kay L. Haarmann
Scott J. Hartzler
JasonC. Head
KandaceA. Heiser
Todd D. Hubal
SusanE. Innes
Paul T. Jakubczak
Ann M. Jellison
Rishi Kapur
Kelly Martin Kingston
John R. Klages, Jr.
ThomasG. Kneer
StevenT. Knight
Brian R. Knox
TanyaM. Kovacevich
Kathryn L. Kritz
Todd J. Kuhl
BrendanM. Leonard
Craig A. Levitz
Xiaoying Liang
Darcy Lindley
Bradley W. Lippowiths
RebeccaM. Locks
Aviva Lubin
JamesI? Lynch
ThomasJ. Macintyre
Richard J. Manship
StephenP.Marsden
JosefE. Martin
RosemaryC. Martin
David M. Maurer
DouglasW. McKenzie
Kirk F. Menanson
DeborahAnn Mergens
Todd A. Michalik
Keith N. Moon
Erica F. Morrone
SharonE. Murray
Maria Nash
David R. Nix
JasonM. Nonis
Randall H. Nordfors
171
JamesL. Norris
Karen A. O’Brien
William T. O’Brien
BarbaraB. O’Connor
Roger D. Odle
Chad M. Ott
Bruce J. Packer
RobertA. Painter
Charles Pare
Carolyn Pasquino
Amy A. Pitruzzello
Phillip A. Pitts
Dylan P. Place
KennethA. Plebanek
Lisa M. Poulin
David N. Prario
Lewis R. Pulliam
John T. Raeihle
Lynellen M. Ramirez
Christopher Randall
Kiran Rasaretnam
Joe Reschini
Mark P. Riegner
Choya A. Robinson
Jeffrey J.
Rozwadowski
Seth A. Ruff
Tracy A. Ryan
Brian C. Ryder
Michelle L. Sands
Christopher P. Sartor
JasonT. Sash
Parr T. Schoolman
Amy V. Shakow
Seth Shenghit
JamesS. Shoenfelt
172
JessieS. Siau
Robin B. Simon
Tracy L. Smith
JesseD. Sornmer
Matthew G. Sorkin
Alan M. Speert
Amy J. Stavros
Laura B. Stein
Michael A. Steinman
CANDIDATES
T. Matthew Steve
Bret L. Stewart
JonathanL. Summers
Brian T. Suzuki
Elizabeth S. Tankersley
JonathanG. Taylor
HuguetteTran
David A. Tritsch
Richard A. Van Dyke
Anil Varma
JayneL. Walczyk
Henry A. Walsh,Jr.
Matthew M. White
Kaylie Wilson
Robert L. Winder
Joel F. Witt
Elissa C. Wolf
Floyd M. Yager
Yuhong Yang
StevenB. Zielke
StevenM. Byam
Kelli R. Caldwell
JanetP.Cappers
Victoria J. Carter
John Celidonio
Yvonne W. Y. Cheng
Aleksandr
Chemyavskiy
Julia F. Chu
Brian K. Ciferri
Kevin M. Cleary
Jeffrey J. Clinch
Nancy J. Collings
Hugo Corbeil
JeffreyA. Courchene
Hall D. Crowder
Claudia M. Barry
Cunniff
JonathanS. Curlee
Amy L. DeHart
Kevin E Downs
Michael E. Doyle
Emilie Drouin
Tammi B. Dulberger
Gregory L. Dunn
Louis Christian Dupuis
Ruchira Dutta
JamesR. Elicker
Brian Elliott
JuanEspadas
Brian A. Evans
Carolyn M.
Falkenstern
Horng-Jiun Fann
Benedick Fidlow
Karen L. Field
Sherri L. Galles
Natalya Gelman
Gary J. Goldsmith
Andrew S. Golfin, Jr.
NatashaC. Gonzalez
Michael J. Grandpre
Martin Halek
Alex A. Hammett
David L. Handschke
Aaron G. Haning
Part 4A
Angela H. A’Zary
Ethan D. Allen
Silvia J. Alvarez
Julie A. Anderson
Satya M. Arya
Roben D. Bachler
Gregory K. Bangs
Amy L. Baranek
Emmanuil Bardis
Brian K. Bell
JamesH. Bennett
HeatherA. Bertellotti
Kristen M. Bessette
Brian A. Bingham
Christopher D. Bohn
Mark E. Bohrer
David R. Border
David C. Brueckman
Paul E. Budde
Marian M. Burkart
Christopher J.
Burkhalter
Brian P Bush
Michelle L. Harnick
Eric C. Hassel
Deborah L. Herman
William N. Herr, Jr.
Cynthia J. Heyer
PeterB. Hindman
Grace Ho
Bradford K. Hoagland
Amy L. Hoffman
Heidi L. Hower
JosephM. Izzo
ChristopherD. Jacks
Karen L. Jiron
Burt D. Jones
Dennis J. Keegan
JeanY. Kim
Kristie L. Klekotka
Matthew R. Kuczwaj
JamesD. Kunce
Brian D. Kurth
Julie-Linda LaForce
Isabelle LaPalme
Douglas H. Lacoss
Mar-tineLaflamme
Robert G. Landau
Brian P. LePage
ChanseoLee
SueJeanLee
BrendanM. Leonard
Eric F. Liland
Chia-Lin C. Liu
JamesR. Lyter
David D. Magee
Alexander P.Maizys
Rajnish Malhotra
RatsamyManoroth
CANDIDATES
StephenJ. McAnena
LawrenceJ.
McTaggartIII
Mea TheodoreMea
Martin Menard
DeborahAnn Mergens
Troy C. Milbrandt
Richard G. Millilo
Ashish Modhera
John Monroe
David P. Moore
Lynn S. Moore
Matthew K. Moran
Lambert Morvan
Michael J. Moss
Malongo Mukenge
RoseMinaMunjee
SusanK. Nichols
John E. Noble
Yanic Nolet
LaureeJ. Nuccio
Jill E. O’Dell
Ann Marie O’Grady
Roger D. Odle
Chad M. Ott
RebeccaW. Palmer
Kelly A. Paluzzi
David J. Persik
Dylan P. Place
Jennifer K. Price
Eileen A. Prunty
Marie-JoseeRacine
Marc Raphael
Amir Rasheed
Jennifer L. Reisig
Neil W. Reiss
173
Michael T. Reitz
Nigel K. Riley
Delia E. Roberts
Kathleen F. Robinson
Efrain Rodriguez
Jennifer L. Rupprecht
ThomasA. Ryan
JosephJ. Sacala
Gary F. Scherer
Jeffery W. Scholl
John R. Scudella
Kelvin B. Sederburg
Ernest C. Segal
Craig S. Sharf
Seth Shenghit
RebeccaL. Simons
Kristen L. Sparks
Michael W. Starke
JonathanC. Stavros
Roxann P. Swenson
ChristopherC.
Swetonic
Michelle M.
Syrotynski
CharlesA. Thayer
Christopher S.
Throckmorton
Craig Tien
Morris Tien
Andy K. Tran
Michael C. Tranfaglia
Laura J. Vargas
Lidia E. Villasenor
Nathan K. Voorhis
Tom C. Wang
Christopher B. Wei
JosephC. Went
ShannonA. Whalen
Patricia C. White
DeanM. Winten
Wendy L. Witmer
John Wong
YuhongYang
Kathermina Lily Yuen
Efren A. Boglio
GenevieveBoileau
Anouk Boucher
Bernard0Bracero,Jr.
RebeccaS. Bredehoeft
Ian R. Brereton
StevenA. Briggs
Mike A. Brisebois
David V. Bruce
ChristopherJ.
Burkhalter
Jennifer A. Burns
Hayden Burrus
Aleksandr A. Bushel
JasonCafaro
JamesM. Campbell
Nathalie Cardinal
SandraL. Carlson
Patrick J. Causgrove
JennieChang
Pei Yi Chao
Hung-WenM. Chen
Lung Chen
Yi-Ju Chen
Chung-WenCheng
Adam K. Cheung
Sheng-ChingChung
Brian K. Ciferri
Derek W. Clayton
Christopher P.Coelho
StevenA. Cohen
Brian M. Collender
Richard JasonCook
PeterJ. Cooper
Brian P.Corrigan
Matthew A. Cowell
SpencerI,. Coyle
JonathanS. Curlee
Dawne L. Davenport
Anne M. DelMastro
ShannonS. Demaree
StephaniePelham
Demeo
Tim P.Deno
Patrick Desrosiers
Kelly D. Dickens
Mark C. Dickey
Marco Dickner
Dean P. Dorman
Paul N. Doss
Richard J. G. Dragon
JoanneAdele Dufault
Gregory L. Dunn
StephanieM. Dupuis
Denis Durand
Marianne E. Dwyer
Cindy Dye
Elaine V. Eagle
Scott W. Edblom
Henry J. Elliot
Part 4B
Cyriaque A. Adou
SajjadAhmad
Giuseppina Alacchi
Ethan D. Allen
Jennifer E. Alper
Scott J. Altstadt
Khurram Amin
John K. Anderson
Maria L. Andrade
Choon-Hong Ang
Amy P.Angel1
Marc D. Archambault
Jennifer D. Arnett
Craig V. Avitabile
Patrick Barbeau
Chad Barber
Emmanuil Bardis
John A. Barnett
Dana Barre
Patrick Beaulieu
John A. Beck
Julie Belanger
Jody J. Bembenek
JamesH. Bennett
Gregory A. Berman
Zahir Bhanji
SarahJ. Billings
Nicole P.Bitros
Cindy M. Bloemer
Michael J. Bluzer
Tanya E. Eng
JonathanPalmerEvans
Chieh Fan
Thomas W. Feltz
Yanjie Feng
StephenM. Finley
Kristine M. Firminhac
StevenA. First
Jeffery A. Fitch
StephenF. Flick
Michael Alexander
Fortier
Antony G. Friel
Timothy J. Friers
Noelle C. Fries
Brian Frost
Martine Gagnon
David E. Gansberg
Michael A. Garcia
GenevieveGaron
Claudine Gauthier
Gregory Evan Gilbert
Jean-FranqoisGirard
Ingrid H. Gokstorp
Alla Golovanevskaya
NatashaC. Gonzalez
Philippe Gosselin
Nicolette A. M.
Goulbourne
StephanieA. Gould
Mathieu Guay
LawrenceA. Guenther
Veronika Hackenjos
David B. Hackworth
Brian D. Haney
Elaine J. Harbus
CANDIDATES
Michael S. Harrington
Sven Hauptfeld
Michael J.
Hawksworth
JasonB. Heissler
Christopher Heppner
Sonia Heroux
PeterB. Hindman
Scott M. Hippen
Jill K. Hoffman
Robert G. Holdom
SusanE. Holland
Allen J. Hope
Ya-Chun Huang
Julie Hubert
Naomi M. Hudetz
Jui-Ruei Hung
Wan M. Idris
JamshaidIslam
StephenA. James
Edward Jhu
JamesJL. Jiang
Michael S. Johnson
TheodoreA. Jones
JeremyM. Jump
Vasilios Kakavetsis
JamesM. Kelly
Michael J. Kenney
DeanW. Kepraios
CameronD.
Kimbrough
Kristie L. Klekotka
JamesD. Kunce
JoanneKurys
Zoe M.Y. Kwok
Martin Labarre
175
Marie-Eve Lachance
Douglas H. Lacoss
Kelly Bryant Lambert
Isabelle Lauziere
Gilles Lavoie
Dennis H. Lawton
Richard G. LeRose
Dominique Lebel
Chang Lee
Norman ShaneLeib
Hubert Lemire
Craig A. Levitz
Yueh-Ying Liao
Ken S. Lim
Lincoln Y.D. Lin
Shiu-Shiung Lin
Jen-Kai Liu
Xin Liu
Anita M. Lo
Tze T. Lo
Kendrick Lombard0
Alison C. Marek
David Laurent Marleau
Clarke E. Marrin
Lori J. Mattusch
Daniel E. Mayost
Jarilyn A. McCartney
PeterB. McCloud
Ian J. McCracken
Lorol Megan
McCrossan
Mark Z. McGill III
DouglasW. McKenzie
Mea TheodoreMea
Anmol Mehra
Jean-FrancoisMichaud
176
Eric J. Mikulaninec
Lisa S. Miolo
TeresaM. Moll
David P.Moore
Isabelle Moreau
Michael J. Morel1
Robert Morency
David K. Morton
JonathanM. Moss
PeterJ. Mulcahey
Paul R. Myers
Claude Nadeau
Jennifer Y. Nei
Kari A. Nicholson
Nina Hemraj Noorali
Eduard A. Nunes
Avital Ohayon
Adeniyi C. Olaiya
Sheri L. Oleshko
CameronJ. Olig
Christopher E. Olson
David A. Ostrowski
Apryle L. Oswald
Gilbert Ouellet
Stella X. Pan
JeromeZ. Pansera
Erika D. Parker
FrangoisPellerin
Sanford E. Penn
Jennifer L. Pepin
Isabelle Perron
Christopher J. Pezalla
Mary G. Phipps
Richard M. Pilotte
Jan L. Pitts
Jordan J. Pitz
CANLIIDATES
Luc Pomerleau
JonathanW. Porter
Marshall E. Posner
Jennifer K. Price
Edward L. Pyle
M. JosephP.
Raaymakers
Scott Rastin
Leslie E. Reed
Kenneth S. Reeves
Sylvain Renaud
Douglas L. Robbins
Kathleen F. Robinson
Ronald J. Robinson
Efrain Rodriguez
NathanW. Root
Richard A.
Rosengarten
Carrie C. Round
Adam S. Rozman
Debra M. Ruocco
Jennifer L. Rupprecht
Mark T. Rutherford
Frederick D. Ryan
JoanneM. Ryan
Mark B. A.
Samlalsingh
Eric G. Sandberg
Myrene Constance
Santos
Todd R. Saulnier
Jennifer D. Schaefer
David W. Scheible
Gary F. Scherer
Rick D. Schnurr
Robin M. Seifert
Lyle S. Semchyshyn
Michael W.
Shackleford
Gregory C. Shane
Hsiao-Chien Shen
Aviva Shneider
John W. Slipp
StevenA. Smith II
EugeneaY. Sohn
JonathanZhan Shan
Song
Hak Hong Soo
GeorgeDennis Sparks
Michael W. Starke
Robert T. Stevenson
Chris Stiefeling
Gina L. Stout
JaymeP. Stubitz
ThomasW. Stus
Javier J. Suarez
John Suder
Joy M. Suh
StephenJ. Talley
Yun Tan
Ming-Hsi Tang
Diane R. Thurston
Beth S. Tropp
ChenghsienTsai
Maurice C.Y. Tsang
Brian K. Turner
Kieh T. Ty
David S. Udall
David Uhland
SusanB. Van Horn
PascalVerrette
Dominique Vezina
Henri L. VichierGuerre
Christine Vigneault
Jennifer A.
VonSchaven
Alexander David
Wallace
DeshengWang
Su-Mei Wang
Kevin E. Weathers
Kelly M. Weber
CANDIDATES
177
Randall Harrison
Weinstock
Ian C. Weir
Avi Weiss
William M. Whitmore
Erica M. Wilson
Karen N. Wolf
TeresaE. Wolownik
Michael L. Yanacheak
Jeng-ShiuYe
Ka YeeYeung
Christine SeungH. Yu
Hung-Chih Yu
Shan-PiYu
Fu-YangYuan
Kathermina Lily Yuen
Carina Zagury
Richard L. Zarnik
Xu Zhang
Yan Zhu
Michele L. Ziegler
John C. Dougherty
ChristopherS. Downey
Kevin F. Downs
StefvanS. Drezek
Sophie Dulude
Mark Kelly Edmunds
JaneEichmann
SarahJ. Fore
Lilane L. Fox
Robert C. Fox
David E. Gansberg
SusanI. Gildea
Olga Golod
PeterS. Gordon
Daniel C. Greer
David J. Gronski
JacquelineL. Gronski
Scott T. Hallworth
Brian T. Hanrahan
Gregory Hansen
Scott E. Haskell
Eric C. Hassel
Cynthia J. Heyer
Margaret M. Hook
Allen J. Hope
RebeccaR. Hunt
Mangyu Hur
Jean-ClaudeJ. Jacob
PatriceJean
Daniel R. Kamen
Robert C. Kane
Chad C. Karls
Hsien-Ming K. Keh
Jill E. Kirby
ThereseA. Klodnicki
Richard S. Krivo
Alexander Krutov
Dar-JenD. Kuo
Jin-Mei J. Lai
William J. Lakins
PeterLatshaw
Michael L. Laufer
Bradley R. Leblond
Doris Lee
Part 5A
Bijoy Anand
Paul D. Anderson
Mario G. Argue110
Afrouz Assadian
Michael J. Bednarick
Michael J. Belfatti
Mariano R. Blanc0
Kofi Boaitey
Raju Bohra
Hobart F. Bond III
Patrice Brassard
Ron Brusky
Hayden Burrus
Brian R. Coleman
Kimberly S. Coles
Margaret E. Conroy
SharonR. Corrigan
David E. Corsi
Tina M. Costantino
David F. Dahl
Nancy K. DeGelleke
Kenneth R. Dipierro
178
I995 EXAMINATIONS-SCCCESSFC’L
Robin R. Lee
Thomas C. Lee
Charles Letourneau
Xiaoyin Li
Smith W. McKee
StephanieJ. Michalik
Catherine E. Moody
Jennifer A. Moseley
RooseveltC. Mosley
Ethan Mowry
John V. Mulhall
Thomas E. Newgarden
Chris M. Norman
Lowell D. Olson
David A. Ostrowski
Donna M. Pinetti
Dylan P.Place
Mary K. Plassmeyer
John F. Powell
Kara L. Raiguel
JacquelineM.
Ramberger
BrendaL. Reddick
Andrew S. Ribaudo
Marie R. Ricciuti
RebeccaJ. Richard
Sophie Robichaud
Richard A.
Rosengarten
Mark B. A.
Samlalsingh
Rachel Samoil
Timothy D. Schutz
Michele Segreti
David G. Shafer
Linda R. Shahmoon
David J. Shaloiko
Kelli D. Shepard-El
RebeccaL. Simons
Donna I,. Sleeth
CANI~IDATES
Duff C. Sorli
Alan M. Speert
Carol A. Stevenson
Thomas Struppeck
Mark Sturm
Adam M. Swartz
Christopher C.
Swetonic
JosephineL. C. Tan
Hung K. Tang
Ming Tang
Colleen A. Timney
Matthew L. Uhoda
David W. Warren
Matthew J. Wasta
Dean M. Winters
Wendy L. Witmer
Jonelle A. Witte
Part 5B
Paul D. Anderson
Craig V. Avitabile
Paul C. Barone
Anna Marie Beaton
Andrew S. Becker
Michael J. Belfatti
Wayne F. Berner
Jonathan E. Blake
Kofi Boaitey
Hobart F. Bond 111
Travis L. Brank
Glen R. Bratty
Laura G. Brill
Ron Brusky
JonathanM. Deutsch
Hugh E. Burgess
<AnthonyM. Di Lapi
Allison F. Carp
Timothy M. DiLellio
Matthew R. Carrier
KennethR. Dipierro
SharonC. Carroll
LOLIISDurocher
Wan Chan
Nathalie Charbonneau Anthony D. Edwards
StephenM. Couzens
JaneEichmann.
Juan Espadas
Michael J. Curcio
Mary KatherineT.
JonathanPalmerEvans
Dardis
William P. Fisanick
SarahJ. Fore
Mark A. Davenport
Mark A. Fretwurst
Willie L. Davis
SergeGagne
PatriciaA. DeoKathy H. Garrigan
Camp0Vuong
Karen L. Greene
Robert A. Grocock
David T. Groff
JacquelineL. Gronski
Alex A. Hammett
Michelle L. Harnick
Guo Harrison
Christopher R. Heim
Kevin B. Held
David E. Heppen
Thomas E. Hettinger
StephenJ. Higgins, Jr.
Luke D. Hodge
Mangyu Hur
JamisonJ. Ihrke
Philip M. Imm
Susan E. Innes
JosephM. Izzo
Christopher D. Jacks
PatriceJean
Derek A. Jones
Jeremy M. Jump
Michael J. Kallan
Daniel R. Kamen
Alexander Kastan
SusanL. Klein
Paul W. Kollner
Linda Kong
Dawna L. Koterman
Beth A. LaChance
Jin-Mei J. Lai
Yin Lawn
Emily C. Lawrance
Dennis H. Lawton
Bradley R. Leblond
Thomas C. Lee
CANDIDATES
JamesP Leise
CharlesLetourneau
RebeccaM. Locks
Richard P Lonardo
William F. Loyd
Kelly A. Lysaght
Atul Malhotra
JosephMarracello
ThomasD. Martin
Claudia A. McCarthy
Allison M. McManus
SarahK. McNairGrove
William E. McWithey
CatherineE. Moody
RooseveltC. Mosley
Kari S. Nelson
Michael D. Neubauer
Khanh K. Nguyen
KathleenC. Odomirok
ChristopherE. Olson
CharlesPare
Moshe C. Pascher
JavanikaPate1
Michael A. Pauletti
Mark Paykin
Harry T. Pearce
Kevin T. Peterson
Anthony G. Phillips
Richard M. Pilotte
CharleneM. Pratt
WarrenT. Printz
PenelopeA. Quiram
Kimberly E. Ragland
Kara L. Raiguel
Frank S. Rau
Andrew S. Ribaudo
179
David C. Riek
Brian P Rucci
Matthew L. Sather
Jennifer A. Scher
Deborah M. Schienvar
Michael F. Schrah
StevenG. Searle
Robin M. Seifert
AnastasiosSerafim
Meyer Shields
Donna K. Siblik
Richard R. Sims
Alan M. Speert
Carol A. Stevenson
StephenJ. Streff
Mark R. Strona
Thomas Struppeck
Brian T. Suzuki
Rachel R. Tallarini
Yuan Yew Tan
Michael J. Tempesta
SteveD. Tews
Jennifer L. Throm
HuguetteTran
Martin Turgeon
Dennis R. Unver
Michael 0. Van Dusen
Matthew J. Wasta
Shu-Mei Wei
Mark S. Wenger
Jeffrey D. White
Miroslaw Wieczorek
Bruce P.Williams
Kirby W. Wisian
Michael J. Yates
Michael G. Young
180
CANDIDATES
Part 6
Jeffrey R. Adcock
Nancy S. Allen
Phillip W. Banet
Kimberly M. Barnett
Karen L. Barrett
Elizabeth F. Bassett
David B. Bassi
Wayne F. Berner
Kevin M. Bingham
StephenD. Blaesing
Barry E. Blodgett
Kimberly Bowen
Tobias E. Bradley
Michael D. Brannon
Cat-yJ. Breese
Linda M. Brockmeier
Lisa A. Brown
Robert F. Brown
Kirsten R. Brumley
Scott T. Bruns
Alan Burns
PamelaA. Burt
Anthony R. Bustillo
JosephG. Cerreta
Daniel G.
Charbonneau
Henry H. Chen
Darrel W. Chvoy
Alfred D. Commodore
David G. Cook
Christopher W. Cooney
Brian C. Cornelison
Matthew D. Corwin
Angela M. Cuonzo
Malcolm H. Curry
Kenneth S. Dailey
CharlesA. Dal
Corobbo
Sheri L. Daubenmier
Jeffrey W. Davis
CatherineL. DePolo
Brian H. Deephouse
Karen D. Derstine
Mark R. Desrochers
Martin W. Draper
SaraP.Drexler
Denis Dubois
RaymondS. Dugue
Kevin M. Dyke
Annette M. Eckhardt
Jennifer R. Ehrenfeld
Dawn E. Elzinga
Todd E. Fansler
Patrick V. Fasciano
Vicki A. Fendley
Mary E. Fleischli
Jeffrey M. Forden
Christian Fournier
Mark R. Frank
Walter H. Fransen
Kay L. Frerk
Mauricio Freyre
Keith E. Friedman
Gary J. Ganci
ThomasP.Gibbons
JamesB. Gilbert
JamesW. Gillette
StewartH. Gleason
Annette J. Goodreau
Mark A. Gorham
Matthew L. Gossell
Mari L. Gray
Monica A. Grill0
M. Harlan Grove
Greg M. Haft
John A. Hagglund
Adam D. Hartman
Gary M. Harvey
Michael B. Hawley
Lisa M. Hawrylak
Jodi J. Healy
Ronald L. Helmeci
Daniel F. Henke
Ronald J. Herrig
ThomasE. Hettinger
Glenn S. Hochler
Daniel L. Hogan, Jr.
Dave R. Holmes
Brett Horoff
Cheng-Chi Huang
Gloria A. Huberman
David D. Hudson
ThomasD. Isensee
Randall A. Jacobson
SuzanneG. James
ChristopherJamroz
Brian J. Janitschke
Walter L. Jedziniak
Philip W. Jeffery
Philip A. Kane IV
Ira M. Kaplan
Robert B . Katzman
Hsien-Ming K. Keh
Mary C. Kellstrom
Scott A. Kelly
William J. Keros
David N. Kightlinger
John H. Kim
Diane L. Kinner
JosephP. Kirley
Omar A. Kitchlew
Wendy A. Knopf
Elina L. Koganski
Thomas F. Krause
Kirk L. Kutch
Andre L’Esperance
SalvatoreT. LaDuca
StevenM. Lacke
Jocelyn Laflamme
Jean-SebastienLagarde
JoseeLambert
Guy Lecours
Kevin A. Lee
Isabelle Lemay
Charles R. Lenz
StevenJ. Lesser
Jennifer M. Levine
John N. Levy
Philip Lew
Hsin-Hui G. Lin
Christina Link
Robb W. Luck
Tai-Kuan Ly
William R. Maag
JasonN. Masch
Bonnie C. Maxie
Laura A. Maxwell
JamesR. Merz
Alison M. Milford
David Molyneux
Lisa J. Moorey
JaniceC. Moskowitz
Matthew S. Mrozek
Karen E. Myers
Vinay Nadkarni
Lowell D. Nelson
Mindy Y. Nguyen
Mihaela L. O’Leary
Kevin J. Olsen
Michael G. Owen
Kathryn A. Owsiany
Dmitry E. Papush
Fanny C. Paz-Prizant
JeremyP.Pecora
Luba Pesis
John S. Peters
Michael W. Phillips
Mitchell S. Pollack
Anthony E. Ptasznik
Patricia A. Pyle
Daniel D. Rath
RaymondJ. Reimer
Melissa K. Ripper
Dennis L.
Rivenburgh,Jr.
JeremyRoberts
DeniseF. Rosen
SandraL. Ross
JosephF. Rosta
PeterA. Royek
Chet JamesRublewski
JasonL. Russ
181
Julie C. Russell
Rome1G. Salam
Elizabeth A. Sander
Manalur S. Sandilya
Cindy R. Schauer
Christine E. Schindler
Michael C. Schmitz
Ia F. Scholdstrom
Craig J. Scukas
Terry M. Seckel
Kevin H. Shang
Scott A. Shapiro
Andrea W. Sherry
Bret C. Shroyer
Katherine R. S. Smith
L. Kevin Smith
Lori A. Snyder
Linda M. Sowter
Caroline B. Spain
TheodoreS. Spitalnick
William G. Stanfield
Christopher M.
Steinbach
Curt A. Stewart
Avivya S. Stohl
Deborah L. Stone
Thomas Struppeck
C. StevenSwalley
Adam M. Swartz
Yuan Yew Tan
Patricia Therrien
Kellie A. Thibodeau
Laura L. Thorne
W. Mont Timmins
Philippe Trahan
ThomasA. Trocchia
182
1995 EXAhllNATlONS-
SII(‘(‘FSSf-I’I
(‘~\NI~Il)~lrf:S
Turgay F. Turnacioglu
Timothy J. Ungashick
Jennifer S. Vincent
Mary Elizabeth Waak
Edward H. Wagner
Benjamin A. Walden
Isabelle T. Wang
Petra L. Wegerich
RobertG. Weinberg
VanessaC. WhitlamJones
David L. Whitley
Jennifer N. Williams
Jerelyn S. Williams
Laura M. Williams
Trevar K. Withers
ShawnaS. Ackerman
StevenD. Armstrong
Timothy W. Atwill
Herbert S. Bibbero
Gary Blumsohn
George P. Bradley
Donna D. Brasley
Margaret A.
Brinkmann
Ward M. Brooks
Elliot R. Burn
Mark W. Callahan
Ralph M. Cellars
Heather L. Chalfant
Dennis K. Chan
Laura R. Claude
PamelaA. Conlin
William F. Costa
Kirsten J. Costello
Christopher G. Cunniff
Joyce A. Dallessio
JeanA. DeSantis
Behram M. Dinshaw
Andrew J. Doll
Norman E. Donelson
John P. Doucette
PeterF. Drogan
Bernard DuPont
Tammy L. Dye
Paul E. Ericksen
JamesG. Evans
Karen M. Fenrich
John R. Ferrara
Kirsten A. Frantom
Jean-PierreGagnon
Eric J. Gesick
John E. Green
StevenA. Green
Lynne M. Halliwell
AlessandreaC.
Handley
Elizabeth E. L. Hansen
Bradley A. Hanson
Robert L.
Harnatkiewicz
Barton W. Hedges
SuzanneE. Henderson
Bettv-Jo Hill
Brandon L. Wolf
Terry C. Wolfe
Kah-Leng Wang
Stephen K. Woodard
Mark I*. Woods
Perry K. Woole>
Rick A. Workman
Robin Zinger
Amy J. Himmelberger
WayneHommes
Marie-JoseeHuard
ThomasA. Huberty
Anthony Iafrate
JosephW. Janzen
F. Judy Jao
Christian Jobidon
Daniel K. Johnson
Kurt J. Johnson
RebeccaA. Kennedy
JoanM. Klucarich
Richard F. Kohan
Mary D. Kroggel
Howard A. Kunst
Andre L’ Esperance
Matthew G. Lange
StevenW. Larson
Lee C. Lloyd
Cara M. Low
JamesM. MacPhee
JamesM. Maher
Maria Mahon
StenhenN. Maratea
Lawrence F. Marcus
Meredith J. Martin
Dee Dee Mays
Charles L. McGuire III
Kelly S. McKeethan
Daniel J. Merk
Claus S. Metzner
Scott M. Miller
Camille Diane
Minogue
Mark J. Moitoso
Robert J. Moser
Raymond D. Muller
Aaron WestNewhoff
John Nissenbaum
Randy S. Nordquist
Mary Beth O’Keefe
TeresaK. Paffenback
Ajay Pahwa
Nicholas H. Pastor
CANDIDATES
William Peter
Mark A. Piske
Dale S. Porfilio
Michael D. Price
Karen L. Queen
Ellen J. Respler
Gregory Riemer
Brad Michael Ritter
Tracey S. Ritter
Bradley H. Rowe
Jean-DenisRoy
David A. Russell
SeanW. Russell
Letitia M. Saylor
Michael B. Schenk
TheodoreR. Shalack
GersonSmith
Carl J. Sornson
Angela Kaye Sparks
Michael J. StewardII
Katie Suljak
StevenJ. Symon
Marianne Teetsel
Trina C. Terne
Edward D. Thomas
Mark L. Thompson
Dom M. Tobey
TheresaA. Turnacioglu
Robert C. Turner, Jr.
JeromeE. Tuttle
Robert J. Vogel
Patricia K. Walker
Erica L. Weida
Geoffrey T. Werner
Wyndel S. White
Gayle L. Wiener
Jeffery M. Zacek
Christian Jobidon
Cara M. Low
Ajay Pahwa
William Peter
Jean-DenisRoy
SeanW. Russell
Katie Suljak
JeromeE. Tuttle
Robert S. Ballmer II
Timothy J. Banick
Philip A. Baum
Tracy L. BrooksSzegda
Mark E. Burgess
DouglasA. Carlone
Daniel G. Carr
Kevin J. Cawley
Galina M. Center
Francis D. Cerasoli
Gary C. K. Cheung
Part 8C
Dennis K. Chan
Bernard DuPont
Jean-PierreGagnon
Betty-Jo Hill
Marie-JoseeHuard
Part 10
Elise M. Ahearn
Rhonda K. Aikens
Jean-Luc E. Allard
Kerry F. Allison
William M. Atkinson
Karen F. Ayres
183
184
1995EXAMINATIONS-SLICCESSFL!I.CANDlDATES
Laura R. Claude
Jo Ellen Cockley
Frank S. Conde
Mary L. Corbett
Catherine Cresswell
David J. Darby
ReneeHelou Davis
Dawn M. DeSousa
Marie-Julie Demers
Lisa Nan Dennison
Robert G. Downs
David L. Drury
Jeffrey Eddinger
Martin A. Epstein
Dianne L. Estrada
Madelyn C. Faggella
Michael A. Falcone
DeniseA. Feder
Daniel J. Flick
Mary K. Gise
Nicholas P Giuntini
Bradley J. Gleason
Ronald E. Glenn
Marc C. Grandisson
Anne G. Greenwalt
StevenJ. Groeschen
William D. Hansen
Christopher L. Harris
Matthew T. Hayden
Patrick C. Jensen
Charles N. Kasmer
Janet S. Katz
Tony J. Kellner
Brian Danforth Kemp
Deborah E. Kenyon
Kevin A. Kesby
Ann L. Kiefer
Michael F. Klein
Terry A. Knull
Gary R. Kratzer
Adam J. Kreuser
David R. Kunze
Mylene J. Labelle
Blair W. Laddusaw
RobertJ. Larson
Paul B. LeStourgeon
John l? Lebens
Marc-Andre Lefebvre
Aaron S. Levine
GeorgeM. Levine
John J. Lewandowski
BarbaraS. Mahoney
Donald E. Manis
Leslie R. Marlo
Kelly J. Mathson
Robert D. McCarthy
David W. McLaughry
Kathleen A.
McMonigle
Robert F. Megens
Madan L. Mittal
Russell E. Moore
Frarqois Morin
Giovanni A. Muzzarelli
Antoine A. Neghaiwi
Victor A. yjakou
Keith R. Nystrom
Marc Freeman
Oberholtzer
ThomasPassante
Abha B. Pate1
Edward F. Peck
WendeA. Pemrick
Anne Marlene Petrides
Mark W. Phillips
JosephW. Pitts
Denis Poirier
On Cheong Poon
Arlie J. Proctor
Mark S. Quigleq
Donald A. Riggins
Douglas S. Rivenburgh
Sallie S. Robinson
Jay Andrew Rosen
John M. Ruane,Jr.
JamesV. Russell
MelodeeJ. Saunders
PeterR. Schwanke
PeterSenak
Rial R. Simons
Keith R. Spalding
DouglasW. Stang
Richard A. Stock
Collin J. Suttie
JeanneE. Swanson
Cynthia J. Traczyk
Patrick N. Tures
PeterS. Valentine
Charles E.
Van Kampen
David B.
Van Koevering
Kenneth R.
Van Laar. Jr.
Mark D. van Zanden
Trent R. Vaughn
JosephW. Wallen
CANDIDATES
Lisa Marie Walsh
John S. Wright
Cheng-ShengP.Wu
185
Claude D. Yoder
Ronald J. Zaleski
Barry C. Zurbuchen
The following candidates were admitted as Fellows and Associates at the CAS Annual Meeting in November 1995 as a result of
their successful completion of the Society requirements in the
May 1995 examinations.
NEW FELLOWS
RhondaK. Aikens
Jean-LucE. Allard
Kerry F. Allison
William M. Atkinson
Karen F. Ayres
Timothy J. Banick
Philip A. Baum
Gary Blumsohn
Donna D. Brasley
Mark E. Burgess
Mark W. Callahan
Kevin J. Cawley
Ralph M. Cellars
Galina M. Center
Francis D. Cerasoli
Laura R. Claude’
Mary L. Corbett
David J. Darby
ReneeHelou Davis
Marie-Julie Demers
Lisa Nan Dennison
John P.Doucette
Paul E. Ericksen
Dianne L. Estrada
Madelyn C. Faggella
Michael A. Falcone
DeniseA. Feder
Mary K. Gise
Bradley J. Gleason
Ronald E. Glenn
Marc C. Grandisson
Anne G. Greenwalt
StevenJ. Groeschen
William D. Hansen
ChristopherL. Harris
Matthew T. Hayden
SuzanneE. Henderson
Anthony Iafrate
Patrick C. Jensen
JanetS. Katz
Tony J. Kellner
Brian Danforth Kemp
DeborahE. Kenyon
Kevin A. Kesby
Michael F. Klein
Terry A. Knull
Adam J. Kreuser
David R. Kunze
Blair W. Laddusaw
Paul B. LeStourgeon
Marc-Andre Lefebvre
Aaron S. Levine
GeorgeM. Levine
John J. Lewandowski
Maria Mahon
BarbaraS. Mahoney
LawrenceF. Marcus
Robert D. McCarthy
KathleenA.
McMonigle
Russell E. Moore
FranCoisMorin
Antoine A. Neghaiwi
John Nissenbaum
Victor A. Njakou
Keith R. Nystrom
Edward F. Peck
WendeA. Pemrick
Mark W. Phillips
JosephW. Pitts
Denis Poirier
On CheongPoon
Arlie J. Proctor
186
1995EXAM lNATIONS—SUCCESSFUL
Mark S. Quigley
Donald A, Riggins
Bradley H. Rowe
John M. Ruane
James V. Russell
Peter Senak
Rial R. Simons
Keith R. Spalding
Douglas W. Stang
Richard A. Stock
Marianne Teetsel
Cynthia J. Traczyk
Patrick N. Tures
Peter S. Valentine
Charles E.
Van Kampen
David B.
Van Koevering
CANDIDATES
Kenneth R.
Van Laar, Jr,
Mark D. van Zanden
Trent R, Vaughn
Lisa Marie Walsh
John S. Wright
Claude D. Yoder
Ronald J. Zaleski
NEW ASSOCIATES
Karen L. Barrett
Lisa A. Bjorkman
Barry E. Blodgett
Tobias E. Bradley 11
Michael D. Brannon
Steven A. Briggs
Pamela A. Burt
Michelle L. Busch
Martin Carrier
Victoria J. Carter
Darrel William Chvoy
Brian C. Cornelison
Angela M. Cuonzo
Claudia M. Barry
Cunniff
Malcolm H. Curry
Charles A.
Dal Corobbo
Dean P. Dorman
Barry P. Drobes
Jeffrey Eddinger
William P. Fisanick
Kay L. Frerk
Gary J. Ganci
Thomas P. Gibbons
Stewart H. Gleason
Annette J. Goodreau
Mark A, Gorham
Monica A. Grillo
Brian D. Haney
Adam D. Hartman
Scott J. Hartzler
Daniel F. Henke
Gloria A. LindenHuberman
David D. Hudson
Randall A. Jacobson
Suzanne G. James
Brian J. Janitschke
Philip W. Jeffery
Michael S. Johnson
Philip A. Kane IV
Ira M. Kaplan
Hsien-Ming K. Keh
Timothy P. Kenefick
Robert W, Kirklin
Therese A. Klodnicki
Salvatore T. LaDuca
William J. Lakins
Josee Lambert
Thomas C. Lee
Isabelle Lemay
Charles R. Lenz
James R. Merz
Dmitry E. Papush
Charles Pare
Brenda L. Reddick
Dennis L.
Rivenburgh, Jr.
Peter A. Royek
Jason L. Russ
Thomas A. Ryan
Marmlur S. Sandilya
Michael C. Schmitz
Craig J. Scukas
Terry M. Seckel
Raleigh R, Skaggs, Jr.
1995 EXAMINATIONS—SUCCESSFUL
L. Kevin Smith
Lori A. Snyder
Thomas Struppeck
Yuan Yew S. Tan
CANDIDATES
Thomas A. Trocchia
Jennifer S. Vincent
Isabelle T, Wang
Jeffrey D. White
187
David L. Whitley
Kirby W, Wlsian
Mark L. Woods
Floyd M. Yager
The following is the list of successful candidates in examinations held in November 1994.
Part 3B
Angela H. A’Zary
Jennifer A. Abner
Sharyn A. Alfers
Keith P. Allen
Mario G. Arguello
David S, Atkinson
Edward H. Balderstone
Emmanuil Bardis
David B. Bassi
Edward L. Bautista
Chad M. Beehler
Ellen A. Berning
Eric D. Besman
John T. Binder
Koti Boaitey
Josee Bolduc
Michael J. Bradley
Glen R. Bratty
Robert J. Brunson
Debra L. Burlingame
Mary L. Cahill
Hong Chen
Ja-Lin Chen
Yvonne W. Y. Cheng
Jonas O. Cho
Seung-Eun Susan Choi
Louise Chung-ChumLam
Ronald V. Clementi
Robert G. Cober
Steven A. Cohen
Larry Kevin Cordee
Hall D. Crowder
Maura K. Curran
Kristin J. Dale
Robert P. Daniel
Loren R. Danielson
Timothy A. Davis
Andrea L. Della Rocco
Alain P. DesChatelets
Paul N. Doss
Julie A. Ekdom
Alana C. Farrell
Kathleen M, Farrell
Solomon C, Feinberg
Sean P. Forbes
Sarah J. Fore
Joseph B, Galbraith
Anne M. Garside
Leslie A. George
Graham S. Gersdorff
Cary W. Ginter
Joseph E. Goldman
Alla Golovanevskaya
Philippe Gosselin
Stephanie A. Gould
Christopher J. Grasso
Amanda Gress
Diane Grieshop
Curtis A. Grosse
Kimberly Baker Hand
David L. Handschke
Chad A. Henemyer
Laurent Holleville
Wayne Homrnes
Hsienwu Hsu
Steven M. Jokerst
Richard B. Jones
Theodore A. Jones
Daniel R. Kamen
Michael A. Kaplan
Alexander Kastan
Brandon D. Keller
John B. Kelly
Joseph P. Kirley
188
1995 EXAMINATIONS-SUCCESSFLrL
JosephE. Kirsits
Jocelyn Laflamme
Jean-SebastienLagace
Valerie Lavoie
Yin Lawn
ChanseoLee
Chiouray Lin
Hsin-Hui G. Lin
StevenR. Lindley
Chia-Lin C. Liu
Jeffrey Y. Liu
JamesW. Mann
Annmay M. Manrique
David E. Marra
Sarah P. Mathes
Daniel E. Mayost
William B. McAlister
Timothy J. McCarthy
Wayne H. McClary
Robert B. McCleish IV
Melissa L.
McDonough
JoanneM. Missry
Rodney S. Morris
Malongo Mukenge
Seth W. Myers
Lauree J. Nuccio
Michael G. Owen
Patrick M. Padalik
SusanM. Pahl
M. Charles Parsons
CANDIDATES
Michael T. Patterson
Michael A. Pauletti
Sylvain Perrier
Julie Perron
JordanJ. Pitz
Paul M. Pleva
Troy J. Pritchett
Patrice Raby
SuzanneM. Rasch
Sylvain Renaud
Andrew S. Ribaudo
Benjamin L. Richards
JosephineT.
Richardson
RhamondaJ. Riggins
Nigel K. Riley
JeremyRoberts
Ronald J. Robinson
Maureen E. Roma
Benjamin G.
Rosenblum
Adam J. Rosowicz
Hanie A. Rowin
FrancesG. Sarrel
Dianne R.
Schwitzgebel
John R. Scudella
StevenG. Searle
Linda R. Shahmoon
Vladimir Shander
Dawn M. Shannon
Aviva Shneider
Matthew R. Sondag
John H. Soutar
Harold L. Spangler,Jr.
KennethW. Stam
David K. Steinhilber
Donald Swofford
Nitin Talwalkar
Robert M. Thomas II
Laura I,. Thorne
Nicole C. Tillyer
Philippe Trahan
Ronald J. Trahan
Andy K. Tran
SalvatoreM. Tucci
Alice M. Underwood
Danielle T. Van Zwet
Kyle J. Vrieze
Ya-FengWang
Jiang WeidongWayne
Brian D. White
Christopher S. Wohletz
Karen N. Wolf
Mihoko Yamazoe
Sharon M. Yao
Mark K. Yasuda
Kristen K. Yates
Jil I,. York
Kenneth Scott Young
TanyaY. Young
189
Part 4A
JasonR. Abrams
Cheryl R. Agina
Amy P.Angel1
David S. Atkinson
JaneL. Attenweiler
Kim M. Basco
Patrick Beaulieu
StephaneBeaulieu
Tony F. Bloemer
Mary Denise Boarman
Bernard0 Bracero,Jr.
Matthew E. Butler
Lisa A. Cabral
Allison F. Carp
Richard J. Castillo
David U. Cho
Andrew K. Chu
Larry Kevin Conlee
M. Elizabeth
Cunningham
John E. Daniel
Loren R. Danielson
Conrad K. Davids
Mari A. Davidson
Jean-Franqois
Desrochers
Ryan M. Diehl
Derek D. Dunnagan
Robert E. Farnam
Alexander
Fernandez,Jr.
BrendaL. Finlen
Sylvain Fortier
Martine Gagnon
Kathy H. Garrigan
Edward R. Garza
Etienne Gingras
StaceyC. Gotham
Amy L. Grbcich
Paul E. Green
Chantal Guillemette
Brian T. Hanrahan
Michael S. Harrington
JamesA. Heer
JamesD. Heidt
ChristopherR. Heim
ChadA. Henemyer
StephenJ. Higgins, Jr.
Kurt D. Hines
MargaretM. Hook
FrancisJ.
Houghton, Jr.
CandaceYolande
Howell
Jodie M. Hyland-Agan
Michael S. Jarmusik
Kelly A. Jensen
Eric D. Johnson
Bryon R. Jones
BrandonD. Keller
Jeff D. Kimble
StevenT. Knight
RobertA. Kranz
IgnaceY. Kuchazik
RichardA. Kutz
RichardV. LaGuarina
FrancoisLacroix
Anh Tu Le
RamonaC. Lee
Jennifer M. Levine
Hsi-yen Lu
Robb W. Luck
KennethW. Macko
JamesW. Mann
David E. Marra
RosemaryC. Martin
Thomas D. Martin
Julie Martineau
RossH. Michehl
Matthew Mignault
RoseL. Miller
David Molyneux
Bilal Musharraf
SethW. Myers
David E. Nicpon
Gregory P Nini
JasonM. Nonis
Nancy E. O’DellWarren
Randall W. Oja
Gilbert Ouellet
Christopher K. Perry
Kevin T. Peterson
ChristopherJ. Pezalla
Richard M. Pilotte
JordanJ. Pitz
KennethA. Plebanek
Judy L. Pool
PenelopeA. Quiram
Kara L. Raiguel
Sylvain Renaud
Jennifer E. Rice
Benjamin G.
Rosenblum
Tracy A. Ryan
Elizabeth A. Sander
Michelle L. Sands
Keith D. Saucier
Raymond G.
Scannapieco
Stuart A. Schweidel
Terri L. Schwomeyer
Ronald L. Smith
John H. Soutar
Alan M. Speert
Gary A. Sudbeck
Beth M. Sweeney
Keith A. Walsh
JonathanG. Taylor
Julie S. Wang
WadeT. Warriner
Eric D. Telhiard
Christian A. Thielman Jiang WeidongWayne
Kelly M. Weber
Laura L. Thorne
Diane R. Thurston
Scott Werfel
John D. Trauffer
Carolyn White
Philip Tso
Arthur S. Whitson
Amy M. Wixon
Brian K. Turner
Matthew L. Uhoda
Scott M. Woomer
SusanB. Van Horn
Christine SeungH. Yu
Danielle T. Van Zwct
Richard 1~.Zarnik
Robert M. VanBrackle Yin Zhang
Karl C. Von Brockdorfi
Part 48
A. Scott Alexander
Marc N. Altschull
Gwendolyn Lilly
Anderson
Carl X. Ashenbrenner
JonathanBalsam
Daniel Bar-Yaacov
Polina Basanskaya
Marc C. Bastien
Michael J. Belfatti
Brian K. Bell
Nicolas P Bergeron
Kristen M. Bessette
Timothy R. Bishop
Timothy S. Bleick
JosephD. Bogdan
Mark E. Bohrer
Thomas G. Bowyer
Erica P. Brown
PeterJ. Brown
Paul E. Budde
SusanK. Bulmer
IMarianM. Burkart
John C. Burkett
Lisa A. Cabral
Brian A. Cameron
JanetP.Cappers
PeggyChan
Jenny N. Chang
PeggyChang
Shu-Ching Chang
ChristopherJ.
Chaplain
Martin Charron
Hui-Chun Chen
Aleksandr
Chernyavskiy
Chia-Ling Chou
JasonT. Clarke
Kevin M. Cleary
Jeffrey J. Clinch
Melissa Clines
DiAnne D. Clous
Robert B. Collins
JamesA. Conley
PamelaA. Connors
SusanD. Cooper
Hugo Corbeil
Matthew D. Corwin
Julie Cousineau
Mari A. Davidson
Robert E. Davis
Annie Derome
Martin Desautels
John T. Devereux
John D. Diffor
Elana L. Doron
StephanieV. Dupuis
TomaszDuski
Chakib Erquizi
Carl R. Fagenbaum
Dana M. Feldman
Gina C. Ferst
Benedick Fidlow
Julia M. Ford
Debbie H. Fu
RosemaryD. Gabriel
JoanneGalinsky
JamesM. Gallagher
Robert F. Geer
Natalya Gelman
Geoffrey W. Gerow
Barry A. Gertschen
Isabelle Gingras
Sanjay Godhwani
Chris D. Goodwin
Glenda J. Granowski
Melanie T. Green
Robert A. Grocock
Chantal Guillemette
Jack F. Gulick
Brian N. Gustafson
Daniel M. Harris
Guo Harrison
Mark A. Hartman
Philip S. Haynes
Kevin B. Held
Chad A. Henemyer
David J. Hennings
Daniel L. Hermetz
Brent L. Hoeppner
JosephH. Hohman
CANDIDATES
Kevin L. Holmes
Eric J. Hornick
David Houle
Derek J. Houle
Jeff S. Howatt
Kuo-Lung Huang
RebeccaR. Hunt
Caleb E. Huntington
Su-FenHwang
Philip M. Imm
Kristin K. Ives
Christopher D. Jacks
JosephW. Janzen
Karen L. Jiron
Carolyn M. Jolly
Burt D. Jones
Michael J. Kallan
JasonE. Kehrberg
Brandon D. Keller
Douglas H.
Kemppainen
JamesR. Kennedy
JeanY. Kim
Anne Marie Klein
David R. Kotick
Christopher Kremer
RachnaKumar
Ting Kwok
ReubenN. Labendz
HuguesLaquerre
John W. Law
John J. Leonard
ThomasE. Leonard
PeggyLeung
Tu-Yi R. Li
Larry J. Lickteig
191
Jeffrey A. Lookkong
RobertA. Macagnano
Lawrence P.
Macdonald
Michael S. Manno
David E. Marra
EmmanuelMatte
Kirk F. Menanson
Troy C. Milbrandt
Kathleen C. Miller
Eric Morin
Ronald T. Nelson
Leo H. L. Ng
Questor K. H. Ng
Mindy Y. Nguyen
Yanic Nolet
JasonM. Nonis
Darci L. Noonan
JamesL. Norris
Randall W. Oja
SergeA. Ouellette
Robin V. Padwa
PamelaS. Pan
Cosimo Pantaleo
Tom Peng
Charles V. Petrizzi
Dylan P.Place
JonathanP. Polon
Michael Porcelli
Devika Prashad
Yuan Qin
Moshe D. Radinsky
Kara L. Raiguel
Christopher Randall
Ronald S. Rees
Mario Richard
Mark P Riegner
Mark E. Robinson
Jaime J. Rosario
Kemp D. Ross
Robert R. Ross
Celine Rouillard
Scott A. Rushing
JosephJ. Sacala
Kim Schaefer
Jeffery W. Scholl
Karl E. Schwehr
Paul S. Serafini
Craig S. Sharf
Boris Shekhter
Junning Shi
MohammadFaisal
Siddiqi
HeatherA. Smith
Meade G. Smith
Matthew R. Sondag
Shoaib Sooti
Jay M. South
Michele L. Spale
Alan M. Speert
Andrew D. Sponsler
Mark R. Strona
Justin B. Struby
JonathanL. Summers
Karrie L. Swanson
EdwardT. Sweeney
ChristopherC.
Swetonic
JonathanG. Taylor
Andy K. Tran
Michael C. Tranfaglia
David L. Treble
StevenJ. Tutewohl
Mark Tynkov
Alice M. Underwood
Rahul Vaidyanath
Carlos M. Vazquez
Leslie A. Vernon
Richard Viesta
Jon S. Walters
JanetL. Wang
Tom C. Wang
Jiang WeidongWavnc
Patricia C. White
Toby A. White
Wendy L. Witmer
.JoelF. Witt
Scott J. Witt
John Wang
Ken Hoong Wong
Whitman Wai Man Wu
Milton F. Yee
Shuang31.Yu
Yin Zhang
Xiaojing Zhao
Vadim Zinkovsky
John C. Burkett
ChristopherJ.
Burkhalter
Donia N. Burris
Aleksandr A. Bushel
Matthew R. Carrier
Milissa D. Carter
BernadetteM. Chvoy
Brian K. Ciferri
SandraCreaney
Andrew S. Dahl
Mujtaba H. Datoo
Jill A. Davis
Timothy A. Davis
Harin A. De Silva
Brian H. Deephouse
Patricia A. DeoCamp0Vuong
SharonD. Devanna
Timothy M. DiLellio
Cynthia Durbin
Rachel Dutil
Ruchira Dutta
Greg J. Engl
Part 5A
Carl X. Ashenbrenner
Kevin J. Bakken
Emmanuil Bardis
JamesV. Barilaro
JamesH. Bennett
JonathanE. Blake
David R. Border
Thomas S. Botsko
Thomas G. Bowyer
Paul E. Budde
Julie Burdick
Hugh E. Burgess
141) ES,4hllN,~TlONS-SUCCESST:CrL
Kristine M. Esposito
JonathanPalmerEvans
Patrick V. Fasciano
Karen L. Field
ChaunceyE.
Fleetivood
SeanP. Forbes
Hugo Fortin
Mark A. Fretwurst
John E. Gaines
BarbaraB.
Glasbrenner
Andrew S. Golfin, Jr.
Lori A. Gordon
Jay C. Gotelaere
Elaine J. Harbus
Christopher R. Heim
Kevin B. Held
Daniel J. Henderson
PeterB. Hindman
Luke D. Hodge
Brett Horoff
C. M. Ali Ishaq
JosephW. Janzen
Kathleen M. Johnson
Jeremy M. Jump
Linda I. Kierenia
Gary G. Kilb
JeanY. Kim
Russell G. Kirsch
Paul W. Kollner
Robin M. LaPrete
W. Scott Lennox
SergeM. Lobanov
Richard P.Lonardo
JasonK. Machtinger
Daniel Patrick Maguire
John T. Maher
JasonA. hlartin
Victor Mata
Daniel E. Mayost
StephenJ. McAnena
GeorgeJ. McCloskey
Michele L. McKay
Michael B. McKnight
Allison M. McManus
SarahK. McNairGrove
Eric Millaire-Morin
Paul D. Miotke
David Molyneux
ChristopherJ.
Monsour
David I? Moore
Kari S. Nelson
Michael D. Neubauer
Kevin J. Olsen
Richard D. Olsen
David J. Otto
RobertA. Painter
Alan M. Pakula
Harry T. Pearce
Robert B. Penwick
Jeffrey J. Pfluger
Richard M. Pilotte
CANDIDATES
193
Jennifer K. Price
Frank S. Rau
Marn Rivelle
Delia E. Roberts
Nathan W. Root
Robert R. Ross
Janelle P. Rotondi
Tracy A. Ryan
Brian C. Ryder
Christy B. Schreck
Darrel W. Senior
Bipin J. Shah
JamesS. Shoenfelt
Bret C. Shroyer
Brian T. Suzuki
Elizabeth S. Tankersley
VarshaA. Tantri
Hugh T. Thai
Jennifer L. Throm
SadhanaTiwari
Beth S. Tropp
Dennis R. Unver
Michael 0. Van Dusen
ClaudeA. Wagner
Edward H. Wagner
Helen R. Wargel
Kevin E. Weathers
DeanA. Westpfahl
Patricia C. White
Kendall P.Williams
GretchenL. Wolfer
Michael G. Young
194
IYYS EXAMINATIONS
Sl’(‘(‘ESStWI.
(‘~\NI)IlMI’ES
Part 5B
JosephJ. Allard
Ethan D. Allen
Gwendolyn Lilly
Anderson
Satya M. Arya
Robert D. Bachler
Amy L. Baranek
John M. Barish
Michael W. Barlow
Victoria A. Beltz
Schnitzer
Mario Bivetti
Jennifer L. Blackmore
Mariano R. Blanc0
Daniel R. Boerboom
Raju Bohra
David R. Border
Julie Burdick
Christopher J.
Burkhalter
Ann Marie L. Cariglia
Scott A. Chaussee
Jamie Cho\+
Brian K. Ciferri
Philip A. Clancey.Jr.
Jeffrey A. Clements
Eric J. Clymer
Sean0. Cooper
Richard S. Crandall
Jonathan S. Curlee
Gregory E. Daggett
Andrew S. Dahl
David F. Dahl
Harin A. De Silva
David A. DeNicola
Brian H. Deephouse
John C. Dougherty
Kevin F. Downs
John A. Duffy
Rachel Dutil
Jeffrey A. Dvinoff
ThomasJ. Dwycr
Kevin M. Dyke
Richard Engelhuber
Kristine M. Esposito
Ellen E. Evans
Alana C. Farrell
Lawrence K. Fink
Ronnie S. Fowler
Robert C. Fox
Timothy J. Friers
Noelle C. Fries
Martine Gagnon
John E. Gaines
Sherri L. C;alles
Barry A. Gertcchen
Olga Golod
NatashaC. Gon/.ale/.
David B. HackLlorth
Barry R. Haine\
Eric C. Hassel
Cynthia J. Heyer
Amy L. Hicks
PeterB. Hindman
Allen J. Hope
Marie-JoseeHuard
Elizabeth J. Hudson
RebeccaR. Hunt
C. M. Ali lshaq
Gregory 0. Jaynes
William R. Johnson
Robert C. Kane
PanayotisN.
Karambelas
Douglay H.
Kemppainen
Omar A. Kitchlew
WendyA. Knopf
David Kodama
Tanya M. Kovacevich
Kathryn 1,. Kritz
Rocky S. Latronica
PeterLatshaw
Doris Lee
Xiaoyin Li
Xiaoying Liang
Timothy D. Logie
VahanA. Mahdasian
Meredith J. Martin
Laura A. Maxwell
Patrick A. McGoldrick
Kirk F. Menanson
Jill M. Merchant
Katherine F.
Messerschmidt
Paul D. Miotke
ChristopherJ.
Monsour
Jennifer A. Moseley
Ethan Mowry
ThomasE. Newgarden
SusanK. Nichols
Brett M. Nunes
Kimberly A. Oaks
Helen S. Oliveto
Richard D. Olsen
Michael A. Onofrietti
GraceA. Orsolino
Chad M. Ott
David J. Otto
M. Charles Parsons
Wendy W. Peng
John M. Pergrossi
Kraig P Peterson
Jennifer K. Price
Brentley J. Radeloff
William D. Rader,Jr.
JacquelineM.
Ramberger
CANDIDATES
195
William J. Raymond
TeresaM. Reis
Mario Richard
RebeccaJ. Richard
NathanW. Root
Richard A.
Rosengarten
Frederick D. Ryan
Ryan D. Schave
Michael Shane
Bintao Shi
Aviva Shneider
RebeccaL. Simons
Christopher S. Strohl
Gary A. Sudbeck
Elizabeth A. Sullivan
Adam M. Swartz
Nitin Talwalkar
Harlan H. Thacker
Joel A. Vaag
JanetK. Vollmert
Nathan K. Voorhis
Claude A. Wagner
Edward H. Wagner
JosephineM. Waldman
VanessaC. WhitlamJones
Kendall P.Williams
TamaraM. Winton
Jeffrey F. Woodcock
Jodi L. Wrede
Jimmy L. Wright
Ruth Zea
Louis M. Brown
Robert F. Brown
Kirsten R. Bmmley
Ron Brusky
Alan Burns
JosephG. Cerreta
Hsiu-Mei Chang
Henry H. Chen
StephenD. Clapp
Michelle Codere
William B. Cody
Sally M. Cohen
David G. Cook
Sheri L. Daubenmier
Jeffrey W. Davis
RaymondV. DeJaco
Dawn M. DeSousa
John D. Deacon
John C. Dougherty
Christopher S. Downey
PeterF. Drogan
David L. Drury
Louis Durocher
Dawn E. Elzinga
Ellen E. Evans
Sylvain Fauchon
Sholom Feldblum
Vicki A. Fendley
John D. Ferraro
Kristine M. Firminhac
Mary E. Fleischli
Sy Foguel
Part 7
Jeffrey R. Adcock
John Scott Alexander
Nathan J. Babcock
Keith M. Barnes
Kimberly M. Barnett
Paul C. Barone
Elizabeth F. Bassett
Michael J. Bednarick
Michael J. Belfatti
Bruce E. Binnig
Lesley R. Bosniack
Edmund L. Bouchie
Kimberly Bowen
Charles Brindamour
Linda M. Brockmeier
Lisa A. Brown
196
IYYS EXAMlNATfONS~SIIC(‘ESSFUL
Jeffrey M. Forden
Christian Fournier
Walter H. Fransen
Bethany L. Fredericks
Jean-PierreGagnon
David E. Gansberg
Kathy H. Garrigan
Lynn A. Gehant
JamesW. Gillette
Moshe D. Goldberg
Karl Goring
Jeffrey S. Goy
Mari L. Gray
Daniel C. Greer
Christopher G. Gross
NasserHadidi
Greg M. Haft
John A. Hagglund
Lynne M. Halliwell
Scott T. Hallworth
AlessandreaC.
Handley
Gerald D. Hanlon
Michael B. Hawley
Jodi J. Healy
David E. Heppen
Ronald J. Herrig
Thomas E. Hettinger
Cynthia J. Heyer
Luke D. Hodge
Daniel L. Hogan, Jr.
Melissa K. Houck
Linda M. Howell
JaneW. Hughes
Mangyu Hur
Brian E. Johnson
CANDIDATES
JamesB. Kahn
John P. Kannon
Chad C. Karls
Anthony N. Katz
Mary C. Kellstrorn
JamesM. Kelly
John H. Kim
Martin T. King
Diane L. Kinner
Brandelyn C. Klenner
Elina L. Koganski
Terri C. Kremenski
Kirk L. Kutch
Andre L’ Esperance
StevenM. Lacke
Timothy J. Landick
StevenW. I,arson
ThomasV. Le
Bradley R. 1,eblond
Guy Lecours
Betty F. Lee
JamesP. I<eise
StevenJ. Lessr~
Philip Lew
Christina 1,ink
Lee C. Lloyd
Cara M. Low
Robb W. Luck
Kyra D. Lynn
William R. Maag
JosephA. Malsky
Betsy F. Maniloff
JosephMarracello
Bonnie C. Maxie
Claudia A. McCarthy
PatriceMcCaulley
Douglas W. McKenzie
Scott A. McPhee
Jennifer Middough
Michael J. Miller
StephenA. Moffett
Catherine E. Moody
Lisa J. Moore]
RooseveltC. Mosleq
Matthew S. Mrozek
Karen E. Myers
Donna M. Nadeau
Vinay Nadkarni
Helen P.Neglia
CatherineA. Neufeld
JamesD. O’Malley
Kevin J. Olsen
David J. Otto
Ajay Pahwa
Erica Partosoedarso
1,isaM. Pawlowski
Jeremy P. Pecora
Trucie L. Pencak
Judy D. Perr
Daniel B. Perry
Luba Pesis
John S. Peters
Michael C. Petersen
Michael W. Phillips
David J. Pochettino
Mitchell S. Pollack
Dale S. Porfilio
Anthony E. Ptasznik
David S. Pugel
Ni Qin-Fcng
Kiran Rasarctnam
RaymondJ. Reimer
Ellen K. Rein
Andrew S. Ribaudo
Cynthia L. Rice
Melissa K. Ripper
Christopher R. Ritter
Dave H. Rodriguez
SandraL. Ross
Jean-DenisRoy
Chet JamesRublewski
David L. Ruhm
DouglasA. Rupp
JoanneE. Russell
ShamaS. Sabade
Rome1G. Salam
Cindy R. Schauer
Timothy D. Schutz
Michele Segreti
JonathanN. Shampo
Michael Shane
Kevin H. Shang
CANDIDATES
Scott A. Shapiro
Jill C. Sidney
Jeffery J. Smith
Kendra BarnesSouth
Caroline B. Spain
TheodoreS. Spitalnick
William G. Stanfield
Christopher M.
Steinbach
Carol A. Stevenson
Curt A. Stewart
Lori E. Stoeberl
DeborahL. Stone
Brian K. Sullivan
RomanSvirsky
Michael J. Tempesta
Mark L. Thompson
Jennifer M. Tornquist
Philippe Trahan
Amy Beth Treciokas
197
JosephD. Tritz
Kai L. Tse
Laura M. Turner
Mary Elizabeth Waak
Benjamin A. Walden
Denise R. Webb
Erica L. Weida
Robert G. Weinberg
Mark S. Wenger
Carol B. Werner
Miroslaw Wieczorek
Jennifer N. Williams
Robin D. Williams
Bonnie S. Wittman
Brandon L. Wolf
BarbaraA. Wolinski
Kah-Leng Wong
Rick A. Workman
Michele N. Yeagley
Part 9
John P.Alltop
Scott C. Anderson
StevenD. Armstrong
Richard J. Babel
Andrea C. Bautista
Douglas S. Benedict
Cynthia A. Bentley
StevenL. Berman
Eric D. Besman
Barry E. Blodgett
Carol A. Blomstrom
Erik R. Bouvin
GeorgeP. Bradley
MargaretA.
Gary C. K. Cheung
Brinkmann
Kuei-Hsia R. Chu
Rita E. Ciccariello
Lisa J. Brubaker
Christopher J. Claus
PamelaJ. Cagney
Anthony E. Cappelletti Kay A. Cleary
Jo Ellen Cockley
DouglasA. Carlone
Michael E. Carpenter Thomas P Conway
Brian C. Comelison
Daniel G. Carr
William F. Costa
Jill C. Cecchini
JoseR. Couret
Julie S. Chadowski
KennethM. Creighton
HeatherL. Chalfant
Jean-Francois
Angela M. Cuonzo
Chalifoux
CharlesA. Dal Corobbo
198
lYY5 EXAhllh’ATIONS-
JeanA. DeSantis
Behram M. Dinshaw
Tammy L. Dye
Jeffrey Eddinger
David M. Elkins
JamesG. Evans
John R. Ferrara
Carole M. Ferrer0
Ginda Kaplan Fisher
Daniel J. Flick
Kay L. Frerk
Margaret Wendy
Germani
Annette J. Goodreau
Elizabeth E. L. Hansen
David S. Harris
Lise A. Hasegawa
Barton W. Hedges
Noel M. Hehr
Betty-Jo Hill
Amy J. Himmelberger
ThomasA. Huberty
Jeffrey R. Hughes
Christian Jobidon
Kurt J. Johnson
JamesW. Jonske
Ira M. Kaplan
Charles N. Kasmer
Lowell J. Keith
StevenA. Kelner
RebeccaA. Kennedy
Michael B. Kessler
Ann L. Kiefer
JosephP. Kilroy
Joan M. Klucarich
Gary R. Kratzer
SII(‘CI:SXFI
II (‘ANIIIDAIES
Brian S. Krick
Edward M. Kuss
Cheung S. Kwan
SalvatoreT. LaDuca
Mylene J. Labelle
JoseeLambert
Matthew G. Lange
Debra K. Larcher
ThomasC. Lee
Julie Lemieux-Roy
DeanneC. Lenhardt
Roland D. Letourneau
Richard S. Light
Shu C. Lin
JamesM. MacPhee
Donald E. Manis
Kelly J. Mnthson
Camley A. Mazloom
HeatherL. McIntosh
Kelly S. McKeethan
David W. McLaughry
Jeffrey A. Mehalic
Brian JamesMelas
Claus S. Metzner
Scott M. Miller
Mark J. Moitoso
KennethB. Morgan, Jr.
Raymond D. Muller
Turhan E. Murguz
Giovanni A. Muzzarelli
Aaron WestNewhoff
Hiep T. Nguyen
Marc Freeman
Oberholtzer
Richard A. Olsen
Milary N. Olson
Charles Pare
ThomasPassante
Nicholas H. Pastor
Miriam E. Perkins
Mark A. Picke
Gregory J. Poirier
Y ves Provencher
Regina M. Puglisi
Cathy A. Puleo
Karen I,. Queen
Kathleen Mary Quinn
Yves Raymond
Natalie J. Rekittke
Ellen J. Respler
Scott Reynolds
Meredith G.
Richardson
Christine R. Ross
JasonL. Russ
ThomasA. Ryan
RajeshV.
Sahasrabuddhe
Manalur S. Sandilya
Linda M. K. Saunders
Michael C. Schmitz
Michael J. Scholl
Arthur J. Schwartz
Craig J. Scukas
Robert D. Share
M. Kate Smith
Lori A. Snyder
Jay M. South
Klayton N. Southwood
1995EXAMINATIONS-SUCCESSFULCANDIDATES
Kevin D. Strous
Thomas Struppeck
JeanneE. Swanson
Rae M. Taylor
Daniel A. Tess
Dom M. Tobey
Michael J. Williams
TheresaA. Turnacioglu Kirby W. Wisian
Robert W. Van Epps
Rita M. Zona
Jennifer S. Vincent
Barry C. Zurbuchen
Kimberley A. Ward
199
NEW FELLOWSADMITTED MAY 1995
I
Front row, from left: Brett E. hllller. Thomas C. Tote. CAS President Man 11. Kaufman, slathiru Lam?. Bradlcy A. Granger, Peter G. \\\‘~ck. Second row:
John V. Van de Water. John F. Rathgeher, David M. Savage. Eduard J. Pulkstenis. Craig \V. Kliethermes. Jeffery J. Scott. Russell Stcmgiser, Timothy J. Cremln.
New Fellows Admitted in hfag 1995 who are not pictured: Suzanne hlartin. Mark Priven. Eileen hl. Sweeney. and Yuan-Yuan Tang.
NEW ASSOCIATES ADMITTED MAY 1995
Front row, from left: Scott A hlarr~n, Dnnwl K Johnw”.
Anthony
L hlan~~rto.
Gal E Kappeler. C&S President
-Wan 11. Kaufman,
Deborah L \IcCrar).
Anne H !&ore,
Camley A hlazloom.
Kelly S hlcKeethdn
Second row: Jean-Ra)mo”d
Kmp\le>.
Jnhn E Green. Cornuell
H 2ldh. Suwn T Grmer.
Charler
R Gnlhot.
Christopher
H Geermg. Juhe K Mper.
John V Hmto”.
Dchr.1 K Laxher.
Grcgop
D Lnr‘her.
fk
J Gewk.
Bnnn S Krlck Third
ran:
Chnrtmn
Jobdon.
Staen A Green. Kenneth B hlorgnn.
Jr, hlarc LaPdIme. Jason h Hoffman.
Kc\ I” T \lurph!.
Da !d S Hanl,.
Sllchacl
K \lcCutchan.
Betty-Jo
HIII. Lowell J Kath. Thomas P Kena
Danlcl E Lent\
3
m
13
R
NEW ASSOCIATESADMITTED MAY 1995
1
Front row. frcm left: L,\.I J Hruh;lb.cr. I’.,mcl,, J ( qnc!.
J’ne t. I%!, Alai \Li. kg;!
( tw;.
(‘A5 I’re\ident
:\llan
\I. Kaufman.
Srcxcn L. Bumnn.
Chrl\tnphcr
G.
JIII C. Cccchlm.
Douglas A. Carlone. Tammy L. Dye. William
F.
Cunmft.
S;lndra L. CarIcy.
Xlarun S. Amtrld.
R~mrnn .Ablan. Second TOW: Hearhcr L. Chnlfant.
Costa. Bchram hl. Dnnshau. Chrlitophcr
J. Cl;lu~. Kc\!”
hl. Brady. Dougla\
J. Bradac. Carol A. Blomstrom.
Stephen C. Dugan. Third row: James L. Bresnahan.
Jrnn-Francon\
Chal1foux.
Prrre Drolct. Chnstopher
R. Allan. Scan R. Devlln. Elllott R. Burn. Steven J. F~nkelste~n. James G. Evans. John T. Glcba. Bruce J.
Bcr~mn.
S. Anden
Encson. (ix)
C.K. Chcung. Andre F. Fwwmc.
3
g
FinI nn\, from left: J~,hn i\. K~,ll~rl\. Llcrc~tlh (i Richardson, Mliw 5. Olson. Sntal~s J. Rrk111Le. (Z,\S I’re\identAllan
Il. Kiaufman. I):I\ 1~1h1. Trr~le.
Cynthia L. VI&II. Chnstina L. Scnnnell. S!c!cn J. Symon. Peter 5 Rauner. Second row: Robert J. Walling III. James L. Sutt~ng, Clnudc Penland IV.
Robert E. Quane Ill, Genevieve P~neau, Thomas Passante. hlichael J. Williams. M. Kate Smith. Hiep T. h’puyen. Third row: John B. Sopkowicz,
Nicholas H. Pastor. hlichael J. Scholl. Steven B. White. Michael J. Sperduto, Eric Vairh. Brian L. Ingle, Daniel A. Tess. Scott Reynolds.
New Associates Admitted in May 1995 who are not pictured: John P. Alltop. K. Athula P. Alwis. Steven D. Armstrong, Corey J. Bilot. Carol A. Blomrtrom.
John T. Bons~gnore. Betsy A. Branapan. Tarn E. Bush, William A. Dowell. Jr., Kimberly J. Drennan. Daniel J. Flick. John F. Huddlcston. LI Hwan Hwang.
hlichnel B. Kessler. Ciq R. Kratzer. Edward A. Lindsay. Richard B. Lord. Tncey L. Matthew. Lynne S. McWithey. Claus S. Sfetzner. Paul W. Mills.
William Peter. Rajesh V. Sahasrahuddhe. Xlanly E. Schafer. Scott D. Spurgat, Scott T. Stclljes. Kevin D. Strous. Joy Y. Takahaqhi. Son T. Tu. Elizabeth R. W~esner.
NEW FELLOWS ADMITTED NOVEMBER
--
--.m
1995
G
1.-
Y
3
Ayrer. Suzanne E. Hendcrwn. hladclrn C. Fa+@la. Second I-OH: Chn\topher L. HJ~IF. Donna D. BrasIr\. John hl. Ruane. Ronald E. Glenn. \farc C. Grandlsxon. Anlhonv Iafmte. Trent R.
Vaughn. John S. U’rqhl. Third row: Janet S. Katz. Rlchard A. Sock. James V. Rucscll. Dens Pomrr. Da;id B Van Keserine. Antome A Nrehaw 1. Fourth row: Chare1s.E. Van Kamoen.
g
NEW ASSOCIATES
ADMITTED
NOVEMBER
1995
OBITUARIES
William J. Hazam
Laura J. Manley
Kenneth L. McIntosh
Robert W. Parlin
James W. Thomas
WILLIAM J. HAZAM
1915-1995
William J. Hazam, a past president of the CAS, died December 29, 1995, at his home in Winter Park, Florida.
Hazam was born in Norwich, Connecticut, on February 14,
1915. He earned a bachelor’s degree from Columbia University
in 1936, and a master’s degree in mathematics from the University of Michigan in 1938.
He was a commissioned officer for the US. Navy and served
in World War II as an aerologist. (In the U.S. Navy, the position of “serologist” was equivalent to those of “meteorologist”
or “weather officer,” terms for similar positions that were used
by other branches of the U. S. military.) According to Hazam’s
long-time friend and colleague, Richard L. Johe, one of the stories that Hazam enjoyed repeating related to a World War II
experience “when he correctly predicted a typhoon! thereby providing sufficient warning to enable our Pacific fleet to scatter and
ride out a dangerous storm. His forecast put him on the Admiral’s staff of favored officers for the rest of the war.”
Immediately after World War II, Hazam spent some time in
Sweden where he starred in a Swedish cowboy movie. In that
movie, his famous one line was, “I tink I go home now,” spoken
with a forced Swedish twang.
206
OBITUARIES
207
While studying for the CAS Examinations from 1947 to 1950,
Hazam served as meteorologist for the American Overseas Airlines. He became an Associate of the Society in 1949, and a Fellow the following year.
Hazam began his actuarial career in 1950 at American Mutual Liability Insurance Company where he worked for 26 years.
His boss was Harold J. Ginsburgh, also a former president of
the CAS. The two worked closely for many years at American
Mutual and shared the same “knowledge, sincerity, and intellectual traits,” Mr. Johe remembered. Johe explained that he,
Messrs. Hazam and Ginsburgh, and others, worked “during an
older, confrontational time in the history of the Stock and Mutual companies; thus, many of Hazam’s contributions to our industry will not be found in the binders of the PCAS, but rather
in the dusty files of minutes of long-forgotten industry committee meetings during which industry issues and endless meetings
demanded much of our energy and attention, leaving little time
for writing actuarial papers.”
Hazam continually demonstrated his dedication to the property/casualty actuarial profession by serving as general chairperson of the CAS Education and Examination Committee for four
years, and serving 18 years on various CAS committees, usually in leadership positions. Those committees include the Committee on Programs, the Constitution Committee, Committee
on Professional Conduct, the Nominating Committee, the Committee on Levels of Certification, and the Textbook Committee.
Hazam served as vice president of the Society in 1967, and as
CAS president in 1968. It was under Hazam’s leadership as president that Article II of the CAS Constitution was changed to
include the words “to promote and maintain high standards of
conduct and competence for the members.”
In 1977, he moved to Virginia Beach to become president of
Physicians Underwriting Company, Inc. He was there only a year
before going back to New England to become an independent
consulting actuary in South Windsor, Connecticut. In 1985, he
retired to Reading, Massachusetts, and moved to central Florida
in 1988 to spend his remaining years in retirement.
Johe remembered Hazam as a “man whose confidence I respected and treasured.” Thomas Murrin, also a past president of
the CAS and a long-time friend of Hazam. said that “Bill was a
great guy, always friendly, pleasant, and gentle-and able to see
the humor in business situations.”
He is survived by his wife, Elaine; three sons. Stephen. John,
and Bruce; daughters Margaret Volpe and Linda Williams; one
sister, Ann Hutchinson; and five grandchildren.
OBITIJARIES
209
LAURA J. MANLEY
1957-1995
Laura J. Manley, an Associate of the Society since 1990, died
November 14, 1995, at her home in Abington, Massachusetts, after a long illness.
Manley was born March 9, 1957, in Palmer, Massachusetts.
She graduated in 1980 from Northeastern University with a bachelor’s degree in mathematics, and began her actuarial career at
Commercial Union Insurance Company in Boston. During her
10 years with Commercial Union, Manley participated in the actuarial rotation program and worked in personal lines pricing,
corporate actuarial financial analysis, and reserving and commercial auto pricing. In 1992, she began working as an associate
actuary for the Automobile Insurers Bureau of Massachusetts
in Boston where she contributed to all aspects of the private
passenger rate filing process and developed a legislative pricing
model for Massachusetts automobile insurance. Even while dealing with a critical illness, Manley remained steadfastly committed
to her work and her continued development as an actuary.
“Laura was an inspiration here at the Automobile Insurers
Bureau of Massachusetts, and she will be greatly missed,” said
Daniel J. Johnston, President of the Bureau. At the time of her
application for membership in the Casualty Actuarial Society,
James E. Fletcher of Commercial Union recommended Manley to the Society because of her “high ethical standards.” David
L. Miller, chief actuary at Commercial Union, said that Manley
“displayed competence and integrity in all areas.” Members of
the CAS Office staff remember Manley as a smiling and cheerful meeting participant who visited the registration desk at CAS
meetings and seminars.
Manley enjoyed traveling, skiing, and gardening. She was also
active in neighborhood functions and youth programs in the town
where she lived. Friends and family of Manley have established
the Laura J. Manley Math and Science Scholarship Fund for seniors graduating from her town’s high school. Contributions can
be made in care of the North Abington Cooperative Bank, North
Abington, Massachusetts.
She is survived by her husband, Andy MacKenzie, parents,
four brothers, a sister, and several nieces and nephews.
OBITUARIES
KENNETH
211
L. MCINTOSH
1995
Kenneth L. McIntosh, an Associate of the Society since 1961,
died on January 22, 1995.
At the time he became an Associate, McIntosh was the manager for the Louisiana Rating and Fire Prevention Bureau in
New Orleans. While working there, McIntosh wrote a Proceedings paper, “Mathematical Limits to the Judgment Factor in Fire
Schedule Rating.” In 1965, McIntosh won the CAS WoodwardFondiller Prize for his Proceeding paper, “A Mathematical Approach to Fire Classification Rates.” He also served for four
years on the CAS Committee on Mathematical Theory of Risk.
In 1968, McIntosh moved to Little Rock, Arkansas, to become a property/casualty actuary at the Arkansas Insurance Department. While working in Little Rock, McIntosh presented discussions of two Proceedings papers: “The Minimum Absolute
Deviation Trend Line,” and “The Credibility of the Pure Premium.”
In 1980, McIntosh moved to Atlanta to become property/casualty actuary at the Georgia Insurance Department. He
retired in 1984 and remained in Atlanta until his death.
ROBERT W. PARLIN
1927-1995
Parlin was born January 15, 1927, in Yunchun, Fukien, China,
where his father was a missionary. After his family returned to
the United States, Parlin received a degree from the University of Minnesota, and served in the United States Navy for two
years.
At the time he became an Associate of the CAS in 1960,
Parlin was working as an actuary for Mutual Service Insurance
Companies in St. Paul, Minnesota. When he became a Fellow of
the Society in 1962, he left the actuarial field to become a research associate at the University of Minnesota, College of Medical Science, Laboratory of Physiological Hygiene Research. Parlin spent five years there focusing on epidemiological research in
coronary heart disease, and published a paper entitled “Death
Rates among Physically Active and Sedentary Employees of the
Railroad Industry,” with five other scientists. That paper appeared in the American Journd of Public Ifealth in 1966.
In 1968, Parlin returned to the property/casualty insurance
field and moved to Adickesalle, Germany, to become an actuary for Neckura Insurance. IHe remained in Germany until
1984, when he moved to Herrliberg, Switzerland, just outside of
Zurich. He remained there until his death on August 26. 1995.
He is survived by his wife, Marianne:
daughter, Cathy.
a son, David; and a
Ol3InJAKIIiS
213
JAMES W. THOMAS
1921-1995
James W. Thomas, a Fellow of the CAS since 1956, died
March 26, 1995, in Wethersfield, Connecticut.
Thomas was born March 13, 1921, in LeMars, Iowa. He graduated from the University of Iowa in Iowa City, and served for
four years as a meteorologist with the 9th Army Air Corps during World War II.
At the time of his Fellowship, Thomas was working at the
Travelers Insurance Company as an assistant actuary. In 1969,
he was promoted to associate actuary. According to his wife, M.
Jane (Armour) Thomas, he enjoyed the many friends he made
while working more than 40 years for Travelers during its “heyday.”
During his years at Travelers, Thomas also served on various CAS committees, including three years on the CAS Finance
Committee.
In 1988, Thomas retired to Wethersfield, Connecticut,
he remained until his death.
where
Thomas is survived by his wife, Jane, and two daughters, Jo
Ellen Thomas and Janet Hartmann. He was predeceased by a
son, John.
214
INDEX TO VOLUME
LXXX11
David G. Hartman-May
15, 1995
Michael A. Walters-November 13, 1995
BALANCING
SHARING
TR~\NSACTION
COSTS fade
RISK I,o,\I)
I
12 I
IN RISK
ARRANGEMENLY
Clive L. Keatinge
BAUT,
23
Tom> R.
Discussion: Risk Loads for Insurers
EXAMINRTIONS-S~ICCESSFLU,
FELDBLUM,
CAhwDI\TF:.s.
.
. . . 78
1995
168
SHOLOM
Author’s Reply to Discussion: Risk Loads for Insurers
FINANCIAL
REPORT
. .
GII-LANAI,
WILLIAM
.
. . . 97
167
.
R.
Discussion: Note on the Gap Between Target and Expected
Underwriting Profit Margins; A
.
.
56
GOGOL,DANIEL F.
Discussion: Unbiased Loss Development Factors
HARTMAN,
DAVID
WILLIAM
. . . 72
G.
Address to New Members-May
HAZAM,
.
15, I995
. .... 1
J.
Obituary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...206
KAUFMAN,
ALLAN
M.
Presidential Address-November
13, I995
. . . . . 126
215
INDEX-CONTINUED
Page
CLIVE L.
KEATINGE,
Paper: Balancing Transaction Costs and Risk Load in Risk
Sharing Arrangements .............................
.23
LAURAJ.
MANLEY,
Obituary..
MCINTOSH,
..20 9
....................................
KENNETH
L.
Obituary........................................211
MINUTES
.7
138
1995 Spring Meeting (May) ..........................
1995 Annual Meeting (November) ....................
NOTEON
THE GAP BETWEEN
UNDERWRITING
TARGET
PROFIT MARGINS;
AND EXPECTED
A
Emilio C. Venezian (November 1987)
Discussion by William R. Gillam ......................
56
OBITUARIES
.206
.209
.211
.212
.213
William J. Hazam ................................
Laura J. Manley .................................
Kenneth L. McIntosh .............................
Robert W. Parlin .................................
James W. Thomas ...............................
PARLIN,
ROBERT
W.
Obituary........................................212
PECK,
EDWARD
F.
Discussion: Simulation Test of Prediction Errors of
Reserve Estimation Techniques; A ....................
LOSS
104
INDEX-CONTINUED
Page
PRNUENTIAI,
AI~I~R~SS-NOVE~II~I:R13, 1995
Allan M. Kaufman
.
I26
REPON‘ 01: THE VICI: PKIISIDllr\;l‘-,~I)hlINIs
lK.,Yl ION
160
Sholom Feldblum (November 1990)
Discussion: Todd R. Bault
Author’s Reply to Discussion: Shalom Feldblum
SIMI.JI.,CION TEST OF PREIXCVON EIwo~~s 01,
ESTI~IKIION TECHNIQUES; A
1.0s~ RI:sI:R\‘I:
James Stanard (May 1985)
Discussion by Edward F. Peck
THOMAS, JAMES
104
W.
Obituary . . . . . . . . . . . . . . . . . . . . . . . . . .
UNBIASED
78
97
. . . ...213
Loss DEVELOPMENT FACTORS
Daniel M. Murphy (November 1994)
Discussion by Daniel F. Gogol, Ph.D. . .
VICF.PRBIDENT-ADMINISTRATION;
WALTERS, MICHAEL
RIVORT or: THI:
72
I60
A.
Address to New Members-November
13. 1995
121
Editorial
Committee,
Proceedings Editors
EUGENE C. CONNELL, Editor-Zn-Chief
DANIEL A. CRIFO
WILLIAM E DOVE
CLAUDIA S. FORDE
ELLEN M. GARDINER
JAMES E GOLZ
RICHARD W. GORVETT
GEORGE M. LEVINE
DAVID L. MILLER, (ex-ojfkio)
REBECCA A. MOODY
ANN E. OVERTURF
DEBBIE SCHWAB

entire volume - Casualty Actuarial Society

Transcription

Similar documents

The Actuary

PowerPoint Template - Actuaries Institute

2014 Annual Report - Casualty Actuarial Society

View - Society of Actuaries

2009 - Department of Mathematics - The University of Texas at Austin

Casualty Actuarial Society Sponsor and Exhibitor Opportunities

the On-Site Program

May, 2015 - Creative Arts Society

Supermarket 7/01 reprint

2011 Annual Report - Child Advocacy Services